One physician organization demonstrates how innovation can result in a substantially better tool for measuring physician productivity.


At a Glance

  • Relative value units (RVUs) were developed by the Centers for Medicare & Medicaid Services (CMS) to establish a value for each physician service categorized by a current procedural terminology code.
  • For some time, RVUs provided an effective tool for measuring physician productivity, until CMS decided to adjust RVU weightings annually, undermining the tool’s comparative effectiveness.
  • A new model developed by the clinical practice management plan serving Stony Brook Medicine University Physicians in Stony Brook, N.Y., provides a means to normalize RVUs from year to year, thereby making RVUs again useful for assessing physician productivity

Gauging physician productivity has long been an imprecise undertaking. In the past, charges for services and/or collection rates and volume served as barometers of productivity. However, these measures often proved inconsistent and faulty. Both individual physicians and medical practices often increased their charges, making comparisons of two time periods difficult. Similarly, different physician services do not always represent equal value, which diminishes the usefulness of volume as a productivity measure.

More recently, organizations found that a more effective approach was to use relative value units (RVUs). These measures have been available since the Centers for Medicare & Medicaid Services (CMS) adopted the resource-based relative value scale in 1992, using RVUs as a basis for payment of physician services. RVUs continue to provide an effective tool for measuring physician productivity today, although this has not consistently been the case. 

An RVU establishes a value for each physician service categorized by current procedural terminology (CPT) codes. This approach initially held promise as a tool for year-on-year comparisons of physician productivity, looking at both similar and different specialties. However, the tool’s modeling potential was severely undermined when CMS decided to annually adjust RVU weightings, often adjusting the RVUs for some CPT codes and not others. These adjustments weaken the comparative value of RVUs for productivity analysis. 

Consider, for example, CMS’s 2009 reweighting of RVUs for evaluation and management (E&M) services, which substantially increased the RVU weighting for these codes, thereby artificially increasing the productivity of physicians in specialties that predominantly perform office visits. As a result, this reweighting could cause a cognitive specialty, such as a family medicine group, to appear comparatively more productive than other physician groups, despite having undergone a loss of physicians and an actual decline in productivity.

The good news is that by applying a straight-forward normalization process, organizations can again effectively use RVUs as a comparative tool not only for gauging productivity, but also for evaluating the effectiveness of rate negotiations and tracking volume trends in clinical services—ultimately providing a much crisper view into a medical practice’s clinical activity.

This normalization process was developed by the clinical practice management plan (CPMP) for Stony Brook Medicine University Physicians, headquartered in Stony Brook, N.Y. (CPMP provides billing, collection, and fiduciary functions for the physician organization.) About three years ago, CPMP set out to restore the comparative value of RVUs by developing a model that normalizes the RVU weightings across different years. In doing so, CPMP created tools that have allowed Stony Brook Medicine to perform multiyear comparative analyses at multiple levels, such as by individual service (CPT code), individual physician, full practice group, payer category, and individual payers and plans. The model CPMP uses to normalize RVUs is described in the sidebar below.

Productivity Analysis

The exhibit below represents a productivity report of clinical services activity at Stony Brook over a three-year period. For contracting purposes, the CPT coding system is broken down into service categories (i.e., anesthesia, E&M, laboratory, medicine, radiology, and surgery). The report shows changes in RVUs within each category from 2008 to 2010. More tellingly, the report shows volume shifts that reveal growth trends, which may either confirm or refute assumptions about the medical practice’s service activity.

For instance, the surgery/integumentary category had historically been considered a small component of services at Stony Brook. Therefore, during contract negotiations, rate increases for these services were not aggressively pursued. However, the report discloses that integumentary services are a greater component of total services than had been believed and that they are growing—from 4.5 percent of total RVUs in 2008 to 4.8 percent of total RVUs in 2010. Based on this finding, the organization decided to emphasize integumentary services, which are a growth category, during rate negotiations rather than focus on a weakening category such as cardiovascular surgery, which the report shows declined from 4 percent of total RVUs in 2008 to 3.3 percent in 2010.

Likewise, the exhibit shows that total E&M services, including the general category and services in specialty areas such as NICU and PICU, constitute about 44 percent of volume, prompting the managed care committee to focus over the next few years on increasing payment per RVU for these services.

By providing a quick snapshot of service shifts from year to year, the numbers also can point to problems that require further analysis. For instance, if RVUs are declining in one category, is it because the service area lacks an appropriate number of physicians or because billing practices are flawed, or is the health system leaving money on the table?

For example, Stony Brook’s spinal surgical services were considered a growing practice, so the RVUs should have been steady or increasing. However, reports showed inconsistent RVUs—sometimes declining, sometimes increasing. This inconsistency was ultimately attributed to the decision by several physicians not to participate in the plans of certain key payers.  

Exhibit 1

Kentros_Exhibit1

Payment per RVU

Normalized RVUs also can be used to analyze payments using a payment-per-RVU analysis. The analysis developed by CPMP uses “bookmatched” charges and payments to evaluate payments for claims posted into the billing system for similar six-month periods for each year of interest. (In bookmatching charges and payments, payments are matched to the specific charges/claims for which the payment was made.) CPMP’s approach, reflected in the exhibits accompanying this article, is to compare charges and payments over six-month periods for three consecutive years, 2008, 2009, and 2010. However, different comparative periods may be identified based upon the needs of the organization. 

In CPMP’s claim sample, claims with dates of service falling outside of the six-month period are eliminated. The remaining claims—the “study claims”—represent those billed in the time period and with dates of service falling within the defined period. This approach ensures that claims are chosen with equal maturation between the six-month periods and does not allow a greater collection period for the older claims, which would cause a higher payment rate for the older claims and distort the comparative value of the analysis. 

After the study claims are chosen, the system is queried to identify all payments received within the six-month period plus the following six-month period (“length of claim maturation” period). Therefore, claims are allowed to mature for six to 12 months. Any payments received for claims after the six to 12 month maturity period are eliminated. To further ensure accuracy, payments also are eliminated for those CPT codes/services without corresponding RVUs, such as J-codes. CPMP also determined that including payment for certain high cost drugs, such as Xiaflex J0775, would result in substantial increases in payment per RVU for Medicare, during a time when increases were approximately 1.5 percent. Once total payments are determined, they are divided by the RVUs associated with the study claims. 

The exhibit below, discussed in greater detail under the heading “Tracking Trends in Clinical Services,” provide examples of reports showing percentage changes in payments per RVU at the physician group level and payments per reimbursement category (i.e., payer group), respectively. (These examples are for illustrative purposes only: Actual payments per RVUs are not shown.)

Exhibit 2

Kentros_Exhibit2

Evaluating the Effectiveness of Rate Negotiations

A payment-per-RVU analysis can be categorized at various levels to evaluate rate negotiations. For example, if according to the contract, the value of E&M services is supposed to increase by 5 percent, that increase should be reflected in the change in payment per RVU for E&M services from one year to another.

Essentially, normalizing RVUs eliminates all variables except for the variance in payment per RVUs between periods. Any variance in payment would be due to a change in the fee schedule, the payer not loading the correct fee schedule, or claims processing issues with the payer.

Such reports can be used to resolve contracting issues with physicians.

For example, a subspecialty group within Stony Brook believed a new contract between the payer and the medical practice that replaced payment based upon charges with a new fee schedule resulted in a loss of up to $400,000 for the physician group. 

The health system acknowledged the new fee schedule would result in lower payments, but not the amount that these physicians anticipated. Once the contract was implemented, an analysis of payment per physician group showed that payment per RVU for this service category declined from $91 per RVU to $82 per RVU, a 9.8 percent decline, resulting in a loss of $100,000—well below the 40 percent decline the physicians had believed would occur. 

Exhibit 3

Kentros_Exhibit3

Tracking Trends in Clinical Services

A payment-per-RVU analysis also can confirm expected outcomes or uncover red flags. The exhibit on page 101 shows the percentage change in payment per RVU over a three-year period for Stony Brook’s 18 clinical departments. From 2009 to 2010, there was a 6.2 percent decrease in payments per RVU in radiology services. In fact, in that same period, the medical practice had agreed to a fee reduction in radiology services at the request of a large payer. So even though there was a reduction in payments per RVU, it was consistent with the contracting.

The examples used here tend to show broad categories, but they can be broken down further. For example, the exhibit on page 104 shows payments per RVU by payer category. The commercial category generally represents payers that do not have a contract with the health system, but are paying according to charges. The negative percentages (albeit decreasing) show a downward trend in payment per RVU for commercial payers as more and more payers are paying not based on charges, but on a percentage of Medicare rates.

These categories can be broken down further into individual payers and separate plans within each payer, such as health maintenance and preferred provider organizations, which provides an even more specific comparison of outcomes against the specific type of contract. A more sophisticated database with more detailed data will drill down to payment per RVU at even the physician level.

Currently, Stony Brook is developing an enhanced service category version of a normalized RVU report that will provide greater specificity by further dissecting the broad CPT categories into more than 250 component subsections. The general E&M category, for instance, will be expanded into office, inpatient, and preventive medicine categories. The cardiovascular category will be divided into bypass and valve replacement subsections. The increased specificity will provide an even more precise way to pinpoint movements in services.

Worth the Extra Step

One criticism of “normalized RVU reporting” is that it is labor intensive and represents an additional cost. Although normalizing data may represent an extra step if a healthcare organization already develops standard RVU reports, we have found the extra work is generally minimal, but extremely useful. At Stony Brook, updating “reference” and “modeling” files used in the normalization process took a single data analyst less than a day to complete. Linking new RVU files and adding new CPT codes every year also requires less than a day’s work.

The question is whether standard reports provide the kind of constructive information that is required for a comprehensive comparison. The results of this analysis suggest that normalized RVU reporting provides a means for comparing productivity, service mix, and payment per RVU reporting that is superior to date-specific RVU reporting.

As payers become more resistant to granting fee increases as payment models transform into performance-based structures built upon bonuses and incentives, such analytic tools will perhaps become even more valuable as a way to gauge the impact of the new models on payment in an environment where payment is increasingly threatened. 


Charles Kentros is director of reimbursement and managed care, Stony Brook Medicine University Physicians, Stony Brook, N.Y., and a member of HFMA’s Metropolitan New York Chapter.

Charles Barbato is a data analyst, clinical practice management plan, Stony Brook Medicine University Physicians, Stony Brook, N.Y. 


sidebar

The Normalization Process

Stony Brook Medicine University Physicians’ clinical practice management plan (CPMP) developed a model that normalizes the relative value units (RVUs) used by the Centers for Medicare & Medicaid Services (CMS) as a basis for physician payment. The RVUs require normalization because they are often recalibrated by CMS, diminishing their value for use in year-to-year comparisons.

CPMP initially established a five-year based period between 2006 through 2010 for comparison. An RVU “reference file” was created that catalogued all current procedural terminology (CPT) codes and their corresponding RVUs that existed during this time period. Newly created CPT codes, as well as their corresponding RVUs published by CMS, are added to the “reference” file annually. From the reference file, CPMP then prepared a “modeling” file that contains a listing of all CPT codes (current and deleted) and the most recent RVU available for each code published by CMS. For any CPT code deleted by the American Medical Association, the most recent available RVU was retained.

The “modeling” file, a sample of which is shown in the exhibit on the opposite page, identifies the last year the CPT code was active, the CPT code itself, a description of the service, the type of service, and the RVU for each code (for both facility and nonfacility services). RVU weightings for unlisted codes were calculated by averaging the weighting of the CPT codes in each code family. In addition, each CPT code that is affected by modifiers 80, 62, and 50 is replicated in the modeling file and the RVUs are adjusted to reflect the cutback associated with modifiers 80 and 62. The RVUs are increased to reflect the impact modifier 50 has on the payment.

The CPT code volume (including modifiers 80, 62, and 50) for the analysis period is then linked to the “modeling” file. This process permits a five-year comparison because RVUs are available for all five years, including RVUs for those codes that have been deleted and/or replaced by other codes. For example, the neonatal and pediatric critical care codes from 99293-99300 were replaced with 99468-99480. 

Using this methodology, RVUs are tabulated at the current RVU weightings for the new neonatal codes in the years that they exist (2010 and forward) as well as for the discontinued codes, in the years they existed (2009 and prior) and at the then-current weightings. This approach eliminates the need to crosswalk deleted codes to replacement codes.

If a longer period of time for comparison is preferred, the “reference file” base year universe of CPT codes can be expanded to include the years of interest. Alternatively, if a shorter period is desired, the CPT date range and claims volume can be modified to only include specific years and/or time frames within a year. For example, CPMP is often asked to compare only the current year with the prior year, or a six-month period with the same six-month period of a prior year or years.

Exhibit

Kentros_Sidebar

 

Publication Date: Thursday, August 01, 2013