Estimating Physician Work

The physician work component of each fee is based on estimates of the physician time to provide the service and the intensity of the physician’s effort. The original work estimates were based on research conducted at Harvard University that was completed in the late 1980s.⁴ Physicians were surveyed about the time and effort to provide each service. Physician panels reviewed and adjusted the survey data to ensure appropriate rank orders of services, so that the more time-consuming and difficult services received higher work values. The panels also examined all of the estimates and made appropriate adjustments to ensure that the ranges of work values and proportions of differences in work values were reasonable.

RUC work recommendations — Over time, most of these Harvard-developed RVUs have been revised to reflect changes in the practice of medicine, and RVUs for new services have been added to the physician fee schedule each year.⁵ CMS relies on the RUC to recommend work values for services that have been identified for review and possible revision as well as values for new services. The RUC uses a process that is similar to the one used in the original Harvard study to develop its recommendations. The RUC has refined the original process to make it more systematic and to incorporate more extensive comparisons across services.

After a service has been identified for review, the RUC asks national medical specialty societies and other health care professional organizations about whether they are interested in developing recommendations. Interested specialties that provide the service survey at least 30 practicing physician members about the identified service. The survey instrument collects detailed information on the work and the direct practice expense resources associated with the service. The instrument includes a vignette that describes a typical patient receiving the service (see text box below); it also lists 15 to 25 other services that may be similar to the service being reviewed.

The respondent may choose to use one or more of these reference services as points of comparison. The respondent estimates the time required to see the patient or perform the service and the corresponding level of effort, which, combined, is termed intraservice work (see text box for factors involved in assessing level of effort). The respondent also estimates the preservice work time, when the physician may review records or consult with colleagues or the patient’s family or prepare for a procedure, and the postservice work time, when the physician may document the service or communicate with others.

The specialties that conducted the survey then present the survey results plus information from specialty expert panels to the RUC. The RUC members question the specialty representatives and compare the survey results with the resources required for similar or related services. Concerns have been raised that the specialties that provide the service may have incentives to inflate the required resources. The RUC believes, however, that the discussion and review by the entire committee ensures a fair and appropriate final recommendation to CMS. The RUC members recognize that, because of the budget neutrality requirement in the fee schedule, an increase to the RVUs of any service causes a corresponding decrease in RVUs for others. This means that payment increases to one specialty may cause decreases to all others.⁶

Accuracy of work values — Although there appears to be widespread agreement that physician surveys and judgment are necessary to establish the work estimates and there is general satisfaction with the estimates in the fee schedule, there are some indications that physicians may not provide reliable estimates of the time required to provide individual services. One study conducted for CMS pointed out discrepancies between the fee schedule descriptions of the various types of evaluation and management services and the typical amount of time the physician spends with the patient.⁷ A study of surgical times found that intraservice time estimates used to develop the work RVUs for certain surgical services were overstated.⁸ The study compared the time estimates with actual operating room logs that were used in hospital scheduling. Although this study was also conducted for CMS, its results have not been systematically incorporated into the process for developing the work RVUs. Overstating physician time would not necessarily affect the accuracy of the fee schedule, if the overstatements were consistent across specialties and services. This is because the fee schedule is based on relative resource differences across services rather than absolute differences. However, both studies indicated that the amount of the overstatements differed across medical specialties and type of service.

Estimating Practice Expense

The original Harvard studies did not estimate practice expense RVUs, so in the RBRVS practice expenses were based on historical physician charges until 1999, when CMS began phasing in resource-based practice expense RVUs. CMS’s method for developing resource-based practice expense RVUs relies on estimates of service-specific, direct expenses, pricing data, and specialty-specific, practice-level expenses.⁹ Because the practice expenses are determined for each service with a “bottom up” approach, the direct and indirect practice expenses are scaled to the aggregate survey data to ensure that the practice expense RVUs do not exceed what is indicated by the survey data.

Enumerating and pricing direct practice resources — The direct practice resources required to provide a service are enumerated through the same surveys and specialty expert panels that the RUC uses to collect the work estimates. (See Table 1, next page, for the direct practice resources of three example services.) The medical specialties that collect these data compile them for presentation to the RUC. The RUC reviews the data and then submits a recommendation to CMS for the direct resources required to provide a reviewed or a new service.¹⁰

TABLE 1

Direct Practice Resources for Three Physician Services.

CMS reviews and may modify the RUC recommendation on direct resources and then attaches a monetary value to each input so they can be converted to RVUs. These values come from a CMS database on the prices of direct expense items. Clinical labor costs per minute, which were updated for the 2002 fee schedule, are derived from Bureau of Labor Statistics surveys and other sources. CMS last updated all supply prices for the 2004 fee schedule with information from manufacturer catalogues and other sources. Equipment prices were updated for the 2005 and 2006 fee schedules. CMS has no stated plans to systematically revise the data, although the Medicare Payment Advisory Commission (MedPAC) and the RUC have suggested establishing an update schedule to ensure the prices are complete and accurate.¹¹

For new types of supplies or equipment, CMS requests pricing data from medical specialties and may receive input from the RUC on the type and quantity of the supply or equipment used. CMS reviews these data, which must include detailed descriptions and pricing information from sources such as catalogue pages, invoices, or Web site pages, before incorporating the prices in its database. MedPAC has expressed particular concern about the database prices for new supplies or equipment.¹² It has recommended that these prices be revised after the supply or equipment has been on the market for a certain number of years, because initial prices may fall with increased competition from additional suppliers. In its proposed rule for 2011, CMS outlined a regular process for considering public requests to change prices for supplies and equipment used in providing physician services. In addition, it proposed using an existing government pricing schedule to establish prices for high-cost ($150 or more) supplies.¹³

Estimating the cost of equipment used to provide a particular service requires assumptions about the equipment purchase price, its useful life, the cost of financing, and the hours the equipment is in use. Equipment purchase price is generally obtained from manufacturers’ catalogues and useful life from published hospital information. CMS has been calculating the cost of equipment based on assumptions that equipment was financed at an interest rate of 11 percent and that it is in use 50 percent of the time the physician’s office is open, although there was no strong evidence to support either assumption. In fact, MedPAC research indicated that physician practices typically used their expensive equipment for the majority of hours their offices were open.¹⁴ In response to concerns that its low equipment use assumption overestimated equipment expense, CMS proposed for the 2010 fee schedule to raise the assumption to 90 percent, but only for diagnostic equipment priced at more than $1million.¹⁵ Subsequent legislation, however, reduced the use assumption to 75 percent.

Allocating indirect practice expenses — Ensuring that the fees appropriately account for the indirect expenses required to maintain a physician practice is complicated because there is no universal standard for allocating indirect practice expenses to individual services. Further, the relationship between indirect and direct practice expenses varies across specialties, and most services are provided by multiple physician specialties.¹⁶ For example, the ratio of indirect to direct expenses for dermatology and ophthalmology are more than 60 percent above the average. Emergency medicine, psychiatry, and anesthesiology have an indirect share that is about 50 percent below the average.

CMS uses survey data to estimate, by specialty, total practice expenses and the direct and indirect expense shares. Indirect expenses are allocated to each service in proportion to its direct expenses plus work.¹⁷ When multiple specialties provide a service, its indirect share is calculated as a utilization-weighted average of the indirect shares for each specialty that provides it. In this way, higher indirect expenses are allocated to services with higher direct expenses or higher work values (or both) and to services predominantly performed by specialties with higher indirect expenses.

Updated practice expense data — Originally, data from the AMA’s Socioeconomic Monitoring System (SMS) were used to estimate aggregate practice expenses by physician specialty. The SMS data reflect practice costs from 1995 through 1999. In response to growing concerns that the SMS data were too old, Congress required CMS to accept specialty-specific survey data to supplement the SMS data. From 2001 through 2006, CMS accepted the supplemental data from 13 specialties. These specialties fielded surveys of their members on practice expenses that met CMS requirements with respect to the survey design, sampling methods, and robustness of the data. For each of these 13 specialties, the supplemental data indicated that practice expenses were higher than those measured by the inflation-adjusted SMS survey data. As a result, the practice expense RVUs of the services predominantly performed by the specialties that submitted supplemental survey data were increased. And, because of the budget neutrality requirement, specialties that did not field supplemental surveys faced reduced practice expense RVUs for their services.

In 2007 and 2008, the AMA, along with medical specialty societies, fielded the Physician Practice Information Survey (PPIS) to collect updated practice expense data. The PPIS data, which CMS purchased, are believed to better represent current practice patterns and costs than the SMS and are collected with a consistent survey instrument.

CMS was satisfied by the response rate achieved with the PPIS, and the AMA adjusted the data to account for nonresponse bias. CMS noted that its goal for the survey was to obtain 100 responses for each specialty and that, for the majority of physician specialties, this goal was achieved. However, the number of surveys that were complete enough to calculate practice expense per hour was much lower. Of the 42 physician specialties, only 23 had 70 or more responses used in calculating the practice expense per hour.¹⁸ Eight specialties had fewer than 50 complete responses.

Incorporation of updated data — CMS planned to substitute the PPIS data for the SMS and the supplemental survey data in calculating the practice expense RVUs for the 2010 fee schedule.¹⁹ However, after reviewing comments on its proposal, CMS indicated that rather than basing the practice expense RVUs completely on the new data, it would implement a transition. The practice expense RVUs will be calculated with decreasing proportions of the historical data and increasing proportions of the PPIS data so that, by 2013, they will be based totally on the 2007–2008 PPIS data. For clinical oncology, however, CMS will continue to use the 2003 supplemental data for oncology drug administration services. This is to comply with the Medicare Prescription Drug, Improvement, and Modernization Act of 2003 (section 303).²⁰

Although the transition to the new data will allow more time for providers to adjust to fee changes, it will also extend the use of fees that are not as accurate as they could be in reflecting practice expenses. If the new data had been used without a transition for the 2010 rates, some specialties, such as cardiology, hematology/oncology, interventional radiology, nuclear medicine, urology, and radiology, would have seen 5 percent to 15 percent reductions in practice expense RVUs. Other specialties like family practice, geriatrics, ophthalmology, and physical medicine, however, would have seen modest increases.²¹

Identif ying Ser vices for Review

MedPAC and the RUC itself have been concerned about the way services in the fee schedule are identified as misvalued and, therefore, candidates for RUC review. Medical specialty societies have identified most of the services. The specialty societies, however, have financial incentives to identify only undervalued services. Indeed, the vast majority of recommended changes from the RUC have been to increase the RVUs of reviewed services (see text box above for recommendations from three five-year reviews).

MedPAC made a series of recommendations to the Secretary of Health and Human Services about how to identify services that may be overvalued in the fee schedule. It recommended establishing “a standing panel of experts to help CMS identify overvalued services and to review recommendations from the RUC.”²² The commission noted several factors that may indicate that the resources required to provide a service may have changed, including alterations in length of stay, site of service, volume, and practice expenses. While recognizing the valuable contributions and role of the RUC, MedPAC stressed the need to examine Medicare claims data for indications of changes in resources and that CMS should lead those efforts.

Rather than establishing an expert panel, CMS encouraged the RUC to identify overvalued services and to evaluate those that had never been reviewed and still had Harvard-valued work RVUs or that had had substantial changes in practice expenses.²³ The RUC’s Five-Year Review Identification Work Group focused on identifying services that were often billed together, were fast-growing, experienced shifts in site of service, or were of high intensity. CMS supported the RUC’s development and use of “compelling evidence standards” for considering a specialty society’s argument that the RVUs are no longer correct.²⁴ Starting from the position that current values are correct, compelling evidence to initiate a review would be documentation in peer-reviewed literature of changes in technique, patient population, site-of-service, or other factors that suggest a change in physician work.

The RUC’s Five-Year Review Identification Work Group focused on identifying services that were often billed together, were fast-growing, experienced shifts in site of service, or were of high intensity.

In its proposed rule for the 2011 fee schedule, CMS lays out how it will comply with the PPACA directive to identify misvalued services.²⁵ In addition to specifying situations that may show that a service is misvalued in the fee schedule, the PPACA suggests that CMS may conduct surveys, collect data in other ways, or conduct analyses to determine whether a service is misvalued and to facilitate the appropriate adjustments to its RVUs. It indicates that services with low relative values that are performed multiple times in a visit may be appropriate for combining for a single payment. The law also stipulates that CMS develop a method for validating the RVUs.

Quality of Data Used to Establish the Fees

CMS and the RUC have worked over the years to systematically update their processes for collecting information on the resources used to provide physician services and for reviewing and adjusting the RVUs. The data on physician practice expenses, in particular, has been substantially revised and improved. In the end, however, most of the detailed information on physician resources must come from physicians, who have a direct economic incentive in how these data affect the fee schedule. Medical specialty societies conduct the service-specific surveys for the RUC process and worked with the AMA to field the PPIS. If they fail to raise questions or provide the information about potentially misvalued services, CMS may not have objective, credible data to use in improving the accuracy of the fee schedule.

CMS has declined to form its own group of experts, as recommended by MedPAC, to provide input on misvalued services and to review RUC recommendations.²⁶ CMS indicated that it would not know how to incorporate information from the group into the rulemaking process and coordinate this group with the RUC. Along with the formation of the RUC’s Five-Year Review Identification Work Group, other changes have been initiated to identify services that are potentially overvalued; together, these may balance the interests of medical specialty societies in identifying only undervalued services. CMS has also lacked the resources to analyze physician data to better inform the fee schedule adjustment process. Because CMS does not have the data, it must ask the AMA to perform any additional analyses of total practice expenses with the PPIS data. The AMA has complied with all requests.

The data underlying the fee schedule are often not robust enough to compare resource use across physician practice arrangements, types of patients, or geographic areas.

The data underlying the fee schedule are often not robust enough to compare resource use across physician practice arrangements, types of patients, or geographic areas. The American College of Cardiology, for example, was concerned that the PPIS data were not representative of cardiology practices because the practice expense per hour calculated from it was much lower than the practice expense per hour calculated from the SMS and from the supplemental data. CMS asked its contractor to evaluate this concern, but it was not able to fully address the questions because of the small number of PPIS responses.²⁷ There also are not enough responses to the specialty-conducted surveys that support the RUC process to evaluate resource use differences across respondents. Such analyses would be particularly instructive as policymakers search for alternative health care delivery models to slow spending growth.

Even though the new data sources are widely acknowledged as improvements over existing ones, they have not been fully incorporated into the process for establishing the RVUs as soon as they become available. The PPIS data will be blended with the SMS data over a four-year period. For physician specialties that had practice expenses derived from SMS or supplemental survey estimates that were higher than estimates from the PPIS data, this blending will keep RVUs from falling as quickly. Specialties that have SMS data with lower practice expense estimates than those derived from the PPIS data will not be able to recoup higher practice expenses as quickly.²⁸ Further, these data will become outdated over time, and there is no plan to repeat the PPIS or develop a new data source.