Inspection of Public Comments: Comments received timely will be available for public inspection as they are received, generally beginning approximately 3 weeks after publication of a document, at the headquarters of the Centers for Medicare & Medicaid Services, 7500 Security Boulevard, Baltimore, Maryland 21244, Monday through Friday of each week from 8:30 a.m. to 4 p.m. To schedule an appointment to view public comments, phone (410) 786-7197.

Copies: To order copies of the Federal Register containing this document, send your request to: New Orders, Superintendent of Documents, P.O. Box 371954, Pittsburgh, PA 15250-7954. Specify the date of the issue requested and enclose a check or money order payable to the Superintendent of Documents, or enclose your Visa or Master Card number and expiration date. Credit card orders can also be placed by calling the order desk at (202) 512-1800 (or toll-free at 1-888-293-6498) or by faxing to (202) 512-2250. The cost for each copy is $9. As an alternative, you can view and photocopy the Federal Register document at most libraries designated as Federal Depository Libraries and at many other public and academic libraries throughout the country that receive the Federal Register .

This Federal Register document is also available from the Federal Register online database through GPO Access, a service of the U.S. Government Printing Office. The web site address is: http://www.access.gpo.gov/nara/index.html.

Information on the physician fee schedule can be found on our homepage. You can access this data by using the following directions:

1. Go to the CMS homepage ( http://www.cms.hhs.gov ).

2. Click on “Medicare.”

3. Click on “Professional/Technical Information.”

4. Select Medicare Payment Systems.

5. Select Physician Fee Schedule.

Or, you can go directly to the Physician Fee Schedule page by typing the following: http://www.cms.hhs.gov/medicare/pfsmain.htm.

To assist readers in referencing sections contained in this preamble, we are providing the following table of contents. Some of the issues discussed in this preamble affect the payment policies but do not require changes to the regulations in the Code of Federal Regulations. Information on the regulation's impact appears throughout the preamble and is not exclusively in section V.

In addition, because of the many organizations and terms to which we refer by acronym in this proposed rule, we are listing these acronyms and their corresponding terms in alphabetical order below:

Since January 1, 1992, Medicare has paid for physicians' services under section 1848 of the Social Security Act (the Act), “Payment for Physicians” Services.” This section provides for three major elements: (1) A fee schedule for the payment of physicians' services; (2) limits on the amounts that nonparticipating physicians can charge beneficiaries; and (3) a sustainable growth rate for the rates of increase in Medicare expenditures for physicians' services. The Act requires that payments under the fee schedule be based on national uniform relative value units (RVUs) based on the resources used in furnishing a service. Section 1848(c) of the Act requires that national RVUs be established for physician work, practice expense, and malpractice expense. Section 1848(c)(2)(B)(ii)(II) of the Act provides that adjustments in RVUs may not cause total physician fee schedule payments to differ by more than $20 million from what they would have been had the adjustments not been made. If adjustments to RVUs cause expenditures to change by more than $20 million, we must make adjustments to preserve budget neutrality.

In the July 2000 proposed rule (65 FR 44177), we listed all of the final rules published through November 1999. In the August 2001 proposed rule (66 FR 40372) we discussed the November 2000 final rule relating to the updates to the RVUs and revisions to payment policies under the physician fee schedule.

In the November 2001 final rule with comment period (66 FR 55246), we revised the policy for resource-based practice expense RVUs; services and supplies incident to a physician's professional service; anesthesia base unit variations; recognition of CPT tracking codes; and nurse practitioners, physician assistants, and clinical nurse specialists performing screening sigmoidoscopies. We also addressed comments received on the June 8, 2001 proposed notice (66 FR 31028) for the 5-year review of work RVUs and finalized these work RVUs. In addition, we acknowledged comments received in response to a discussion of modifier-62, which is used to report the work of co-surgeons. The November 2001 final rule also updated the list of services that are subject to the physician self-referral prohibitions in order to reflect CPT and Healthcare Common Procedure Coding System (HCPCS) code changes that were effective January 1, 2002. All these revisions ensure that our payment systems are updated to reflect changes in medical practice and the relative value of services.

The Medicare, Medicaid, and State Child Health Insurance Program (SCHIP) Benefits Improvement and Protection Act of 2000 (Pub. L. 106-554) (BIPA) modernized the mammography screening benefit and authorized payment under the physician fee schedule effective January 1, 2002. It provided for biennial screening pelvic examinations for certain beneficiaries and expanded coverage for screening colonoscopies to all beneficiaries effective July 1, 2001. It provided for annual glaucoma screenings for high-risk beneficiaries and established coverage for medical nutrition therapy services for certain beneficiaries effective January 1, 2002. It expanded payment for telehealth services effective October 1, 2001; required certain Indian Health Service providers to be paid for some services under the physician fee schedule effective July 1, 2001; and revised the payment for certain physician pathology services effective January 1, 2001. This final rule conformed our regulations to reflect these statutory provisions.

The final rule also announced the calendar year 2002 physician fee schedule conversion factor of $36.1992.

This proposed rule would affect the regulations set forth at part 410, Supplementary medical insurance (SMI) benefits and part 414, Payment for Part B medical and other health services.

Section 121 of the Social Security Act Amendments of 1994 (Pub. L. 103-432), enacted on October 31, 1994, required us to develop a methodology for a resource-based system for determining practice expense RVUs for each physician's service beginning in 1998. In developing the methodology, we were to consider the staff, equipment, and supplies used in providing medical and surgical services in various settings. The legislation specifically required that, in implementing the new system of practice expense RVUs, we apply the same budget-neutrality provisions that we apply to other adjustments under the physician fee schedule.

Section 4505(a) of the Balanced Budget Act of 1997 (BBA) (Pub. L. 105-33), enacted on August 5, 1997, amended section 1848(c)(2)(ii) of the Act and delayed the effective date of the resource-based practice expense RVU system until January 1, 1999. In addition, section 4505(b) of the BBA provided for a 4-year transition period from charge-based practice expense RVUs to resource-based RVUs.

Further legislation affecting resource-based practice expense RVUs was included in the Medicare, Medicaid and State Child Health Insurance Program (SCHIP) Balanced Budget Refinement Act of 1999 (BBRA) (Pub. L. 106-113), enacted on November 29, 1999. Section 212 of the BBRA amended section 1848(c)(2)(ii) of the Act by directing us to establish a process under which we accept and use, to the maximum extent practicable and consistent with sound data practices, data collected or developed by entities and organizations. These data would supplement the data we normally collect in determining the practice expense component of the physician fee schedule for payments in CY 2001 and CY 2002. (In the 1999 final rule (64 FR 59380), we extended, for an additional 2 years, the period during which we would accept supplementary data.)

Effective with services furnished on or after January 1, 1999, we established a new methodology for computing resource-based practice expense RVUs that used the two significant sources of actual practice expense data we have available—the Clinical Practice Expert Panel (CPEP) data and the American Medical Association's (AMA) Socioeconomic Monitoring System (SMS) data. The methodology was based on an assumption that current aggregate specialty practice costs are a reasonable way to establish initial estimates of relative resource costs for physicians' services across specialties. The methodology allocated these aggregate specialty practice costs to specific procedures and, thus, can be seen as a “top-down” approach.

A brief discussion of the major steps involved in the determination of the practice expense RVUs follows. (Please see the November 1, 2001 final rule (66 FR 55249) for a more detailed explanation of the top-down methodology.)

• Step 1 —Determine the specialty specific practice expense per hour of physician direct patient care. We used the AMA's SMS survey of actual aggregate cost data by specialty to determine the practice expenses per hour for each specialty. We calculated the practice expenses per hour for the specialty by dividing the aggregate practice expenses for the specialty by the total number of hours spent in patient care activities.

• Step 2 —Create a specialty specific practice expense pool of practice expense costs for treating Medicare patients. To calculate the total number of hours spent treating Medicare patients for each specialty, we used the physician time assigned to each procedure code and the Medicare utilization data. We then calculated the specialty specific practice expense pools by multiplying the specialty practice expenses per hour by the total physician hours.

• Step 3 —Allocate the specialty specific practice expense pool to the specific services performed by each specialty. For each specialty, we divided the practice expense pool into two groups based on whether direct or indirect costs were involved and used a different allocation basis for each group.

(i) Direct costs—For direct costs (which include clinical labor, medical supplies, and medical equipment), we used the procedure specific CPEP data on the staff time, supplies, and equipment as the allocation basis.

(ii) Indirect costs—To allocate the cost pools for indirect costs, including administrative labor, office expenses, and all other expenses, we used the total direct costs combined with the physician fee schedule work RVUs. We converted the work RVUs to dollars using the Medicare CF (expressed in 1995 dollars for consistency with the SMS survey years).

• Step 4 —For procedures performed by more than one specialty, the final procedure code allocation was a weighted average of allocations for the specialties that perform the procedure, with the weights being the frequency with which each specialty performs the procedure on Medicare patients.

(i) Zero Physician Work Pool—For services with physician work RVUs equal to zero (including those services with a technical and professional component), we created a separate practice expense pool using the average clinical staff time from the CPEP data and the “all physicians” practice expense per hour.

We then used the adjusted 1998 practice expense RVUs to allocate this pool to each service. Also, for all radiology services that are assigned physician work RVUs, we used the adjusted 1998 practice expense RVUs for radiology services as an interim measure to allocate the direct practice expense cost pool for radiology.

(ii) Crosswalks for Specialties without Practice Expense Survey Data

Since many specialties identified in our claims data did not correspond exactly to the specialties included in the SMS survey data, it was necessary to crosswalk these specialties to the most appropriate SMS specialty.

(iii) Because we believe that most physical therapy services furnished in physicians' offices are performed by physical therapists, we crosswalked all utilization for therapy services in the CPT 97000 series to the physical and occupational therapy practice expense pool.

Rank order anomalies were created in three ophthalmology families of codes because only certain services in each family were brought to the Practice Expense Advisory Committee (PEAC) for refinement, while CPEP data for the other codes were left unchanged. The American Academy of Ophthalmology has requested that we make the following changes in the CPEP data to ensure that the more complex services in a family of codes are not paid less than the simpler services and we are proposing to do so.

CPT code 67820, Revise eyelashes—remove ophthane from the supply list.

CPT code 67825, Revise eyelashes—remove the bipolar handpiece from the supply list.

CPT code 65220, Removal foreign body from eye—use the supply list and clinical staff time assigned to CPT code 65222. The exam lane should be the only equipment assigned.

CPT codes 92081 and 92083, Visual field examination(s)—Assign the same supplies and equipment as CPT code 92082; assign 35 minutes of clinical staff time to 92081 and 70 minutes to 92083.

There are three CPT codes for transurethral destruction of prostate tissue: CPT 53850, by microwave therapy, CPT 53852, by radiofrequency thermotherapy, and CPT 53853, by water-induced thermotherapy (WIT). A manufacturer of WIT equipment has expressed concern that the practice expense inputs currently assigned to CPT 53853 underestimate the costs associated with that procedure relative to the other two codes. We have compared the inputs of the three codes and agree that the WIT procedure has not been assigned many of the basic supply and equipment inputs that are included in the CPEP inputs for the other two procedures. Therefore, we are proposing to add, on an interim basis, the following inputs: Power table, ultrasound unit, mayo stand, endoscopy stretcher, light source, chux, sani-wipe, patient education book, sterile towel, sterile gloves, specimen cup, alcohol swab, gauze, tape, lidocaine, betadine, 10 cc syringe, 30 cc syringe, sterile water, leg bag. These inputs would be in addition to the thermotherapy unit, treatment catheter, drainage bag, cystopak (which contains drapes, syringe, irrigation tubing, surgical lubricant, sterile cup, gauze pad and cysto tubing) and minimum visit package for each visit (which contains patient gown, unsterile gloves, exam table paper, pillow case and thermometer probe cover) that are currently assigned as practice expense inputs for this procedure.

We are also proposing to change on an interim basis the staff type for CPT code 53853 from the RN/LPN/MTA blend to RN in order to make the staff type consistent among these three similar procedures. In addition, we have corrected for all three procedures the minutes assigned to each piece of equipment to reflect the intra- and post-clinical staff times only, rather than the total clinical staff times as we do for all services.

We will request that these three procedures be reexamined by the PEAC at the same time in order to ensure that there is a consistent approach to the assignment of direct cost inputs.

We have also received questions regarding the large disparity in prices that we have used for the three different thermotherapy machines. Currently, the thermotherapy equipment for CPT code 53850 is priced at $180,000, code 53852 at $42,995 and code 53853 at $18,500. The first two prices were given to us in 1999 and we have been sent documentation that indicates that the prices have fallen dramatically since that time. This documentation indicates that the current price for the thermotherapy equipment for CPT 53580 is somewhere between $55,000 and $100,000 and for code 53852 between $25,000 and $30,000. We are proposing to set the prices at $60,000 and $30,000, respectively and are also requesting that commenters furnish any additional available price documentation, particularly invoices for recently purchased equipment, so that we can ensure that any differences in assigned prices accurately reflect actual differences in costs.

It has been brought to our attention that the electrodes assigned to the supply list for CPT code 97033, Iontophoresis, are not the type of electrodes required for this procedure. We are proposing to substitute two electrodes with a medication vesicle as the appropriate supply for iontophoresis.

The current price of $40.00 for 1000 ml of sterile water that is listed in our CPEP supply database is incorrect. We are proposing to change this to $3.00.

Within the last year, there have been two reports that have addressed the zero physician work pool. The GAO released a report in October 2001 that recommended eliminating the zero physician work pool (GAO-02-53, page 25). The Lewin Group, under contract with us, provided a draft report on June 5, 2001 analyzing the zero physician work pool and provided several ideas that we could study as alternatives. As we indicated in the November 2, 1998 final rule (63 FR 58821), we created the zero physician work pool as an interim measure until we could further analyze the effect of the top-down methodology on the Medicare payment for services that do not have physician work RVUs. Given our interest in finding alternatives to the zero physician work pool, we have analyzed the following possible ideas:

• Eliminate the Zero Physician Work Pool.

The Lewin Group indicated that one idea could be to eliminate the zero physician work pool, as recommended by the GAO (the Lewin Group, page 19). We do not believe that the zero physician work pool should be eliminated at this time. In the absence of a change to the methodology or additional data, eliminating the zero physician work pool would result in large reductions in payments for some of the specialties whose services are included in the pool. The Lewin Group also indicated that this idea is not a “viable alternative to the current zero (physician) work pool approach.” (The Lewin Group, pages 19-20).

• Develop Specialty-Specific Zero Physician Work Pools.

Under this approach, the Lewin Group report described an idea for maintaining the general zero physician work pool approach with specialty-specific zero physician work pools (the Lewin Group, page 20). Since the zero physician work pool is an exception to the basic methodology, we are not currently interested in developing another exception to replace it. We are interested in finding a single methodology that would apply to all physicians' services. While we are not adopting this suggestion, we do appreciate the Lewin Group's work in developing this and many other ideas for consideration as we make refinements to the practice expense methodology.

• Make Technical Component Equal Global Less Professional Component RVUs.

Many of the services that are affected by the zero physician work pool are services that have both professional and technical components. Under current policy, the technical component practice expense RVU is determined in the zero physician work pool. It is added to the practice expense RVUs for the professional component determined under the basic methodology to determine payment for the global service. This Lewin Group idea would change this to make the technical component RVUs equal the difference between the global and the professional component RVUs while other zero physician work services would be returned to the basic methodology (The Lewin Group, page 21).

If we were to adopt this approach, the zero physician work pool would no longer have any effect. The zero physician work pool would not have any effect on the professional and technical component services since the global service from the basic methodology would be used to derive the technical component value. The practice expense RVUs for other zero physician work services would be priced under the basic methodology. In the absence of a change to the methodology or additional data, this idea would result in large reductions in payments for some of the specialties whose services are included in the pool.

As we have indicated above, we are concerned about adopting this idea at this time. While we are not currently proposing to adopt this idea as an alternative to the zero physician work pool, we do believe there is merit in making the technical component value equal the difference between the global and professional component RVU for services that are unaffected by the zero physician work pool. We receive many more bills for the global than the technical component only. Since it is far more common to receive a global than a technical component only bill, it is far more likely that using the global to value the technical component service will result in a payment that is more typical of the relative actual practice expense associated with the service.

For this reason, we are proposing to make the technical component value equal to the difference between the global and the professional component for procedure codes that are not included in the zero physician work pool. We will continue to make the global value equal to the sum of the professional and the technical component values for procedure codes that remain in the zero physician work pool. However, we may revisit this decision in the future if we can address issues related to the zero physician work pool. We have provided more detail on the redistributive impact of this proposal among all physician specialties in the impact section of this proposed rule.

• Develop Proxy Physician Work RVUs for Zero Physician Work Services.

Finally, the Lewin Group described an idea that would retain work and direct expenses as the basic allocators of indirect costs but create proxy physician work values for services that have no physician work (the Lewin Group, pages 22-23). The GAO suggests that the basic method for allocating indirect expenses does not adequately account for the indirect costs associated with services that do not have physician work RVUs (GAO-02-53, page 22). We do not believe that the large payment reductions that would occur if some zero physician work services were priced under the basic methodology are necessarily associated with the indirect cost allocation methodology. While the zero physician work pool is of benefit to many of the services that were originally included, some specialties commented that this methodological change negatively affected the particular services they provide. As a result, we allowed specialties to request that their services be removed from the zero physician work pool (see July 22, 1999 proposed rule (64 FR 39620)). If there are shortcomings in the indirect cost allocation for services that have zero physician work, it seems likely that the values for all zero physician work services would be adversely affected. However, since many zero physician work services are not adversely affected under the top down methodology, it seems unlikely that the indirect cost allocation explains the adverse payment impact that would result for some services from elimination of the zero physician work pool. For this reason, we do not anticipate modifying the indirect cost allocation for zero physician work services.

Based on our analysis of the Lewin suggestions, we do not believe that allocation of either direct or indirect expenses explains the effect of the top down methodology on zero physician work services. Rather, we believe it is likely that a relatively low practice expense per hour for some of the specialties included in the zero physician work pool explains why their payments are adversely affected by its elimination.

The specialties whose services are affected by the zero physician work pool may want to conduct supplemental practice expense surveys if they believe there are shortcomings in the practice expense per hour information that we use as part of the basic methodology. We have published in this issue of the Federal Register, an interim final rule with comment that will modify the criteria for acceptance of supplemental data. This should make it easier for specialties to incorporate new practice expense survey information into the methodology. Further, as we indicated previously in the November 1, 2000 Federal Register (65 FR 65384), we believe that there are significant advantages to receiving practice cost information through multi-specialty surveys. For this reason, we would welcome a multi-specialty practice expense survey from all of the specialties that have payments affected by the zero physician work pool.

In the June 5, 1998 proposed rule (63 FR 30832), we proposed an adjustment to the medical supplies practice expense per hour for oncology as a result of a concern that their inordinately high practice expense per hour included expenses associated with separately payable cancer drugs. We proposed to substitute the “all physician” average for the oncology-specific medical supplies practice expense per hour. We received public comments indicating that, even after excluding the effect of higher drug expenses, oncologists have higher medical supply expenses than the average physician because of high supply costs associated with the administration of chemotherapy. These commenters suggested alternatives to using the average physician rate. In our November 2, 1998 (63 FR 58825) final rule, we made an adjustment to the medical supplies practice expense per hour for oncology and indicated our belief that oncology medical supply expenses would not necessarily exceed those of the average physician. However, the adjustment has largely had no effect since the practice expense RVUs for chemotherapy administration services are determined in the zero physician work pool.

In its October 2001 report, the GAO recommended that we examine the effect of the adjustment made to oncologists' reported medical supplies expenses per hour. GAO did not suggest a specific alternative to the adjustment we made (GAO-02-53, pages 24-25). Consistent with the GAO recommendation, we have examined this adjustment and its impact on Medicare payments to oncologists. Upon further review, we believe that there is merit in reconsidering the adjustment that we made to the medical supply expenses for oncologists in combination with removing chemotherapy administration services from the zero physician work pool.

At this time, we have no specific information on oncology medical supply expenses net of separately payable drugs. However, we have established a process that would allow specialties to submit supplemental practice expense survey data to us. While the criteria for performing a survey require consistency with the SMS, we are amenable to modifications to the survey instrument so that it can address questions that are of concern to a specific medical specialty. For instance, we would allow an oncology survey to request that respondents distinguish between drug and other medical supply related expenses. We believe that using specific data on this question from a survey would be preferable to developing an alternative adjustment that requires us to make assumptions about oncology medical supply expenses. However, if further survey information is unavailable to us, we are considering information that could be used as a reasonable proxy to determine the portion of the supplies practice expense per hour that is attributable to medical supplies that are not separately payable. Such an idea was suggested in the public comments on the June 5, 1998 proposed rule. We are considering other alternatives as well. These approaches to the supplies practice expense per hour would apply if chemotherapy administration services were removed from the zero physician work pool.

In the November 2, 1998 final rule (63 FR 58841), we indicated that average clinical staff time was used in the creation of the zero physician work pool. Since the cost pools are created based on physician time and, by definition, zero physician work services have no physician time, we need to use staff time to create the cost pool. If our database indicates that multiple staff types are typically involved in the service, we have used an average of the different clinical staff times. We are proposing to create the cost pool using the highest staff time in place of average staff time. The impact of this proposal is shown in the impact section of this proposed rule.

We are proposing to remove the non-invasive vascular diagnostic study codes (CPT codes 93875-93990) from the zero physician work pool based on a request from the American Association for Vascular Surgery and the Society for Vascular Surgery. The impact of this proposal is also described further in the impact section.

As discussed above, in the November 2, 1998 final rule (63 FR 58841), in response to the many commenters who were concerned about the proposed reductions for services with zero physician work RVUs, we created a separate practice expense pool for all services with zero physician work RVUs. The assignment of services to this zero physician work pool was of benefit to most services in this expense pool. However, some specialties were negatively affected by this methodology, and we have allowed specialties to indicate whether their services should be priced in this pool.

Immunization administration services do not have physician work RVUs and have been included in the zero physician work pool. So that the direct practice expense resource costs associated with the immunization administration services are recognized, we propose removing these services from the zero physician work pool methodology and treating them like the vast majority of services on the physician fee schedule. Using the direct cost practice expense inputs as recommended by the AMA's RUC, the proposed practice expense relative value units will be 0.22 for CPT code 90471 and 0.09 for CPT code 90472. This change will nearly double payment for CPT code 90471 and slightly reduce payment for CPT code 90472. Procedure CPT code 90471 is used for immunization administration and CPT code 90472 is used for each additional vaccine. Since CPT code 90472 must be billed in conjunction with CPT code 90471, the total payment for these procedures will increase when billed together.

We have not assigned immunization administration physician work RVUs because this service does not typically involve a physician. The nurse that administers the vaccine typically provides the necessary counseling to the patient and this time is accounted for in the practice expense RVU.

In addition, we would note that not all services represented by CPT codes 90471 and 90472 are covered by Medicare. For example, medically necessary administrations of tetanus toxoid (such as following a severe injury) would be covered whereas preventive administration of this vaccine would not be covered. Also, we will consider whether the amount of counseling of the patient and/or family may be different for childhood immunizations than for the typical Medicare service. Therefore, we are considering whether coding changes to reflect these differences would be appropriate.

As indicated earlier, Medicare utilization is an important data source used in determining the practice expense RVUs. In our final rule published on November 2, 1998 (63 FR 58815), we used 1997 Medicare utilization data to create the original resource-based practice expense RVUs. Based on a public comment, we indicated in our November 2, 1999 final rule (64 FR 59405) that we would use 1998 Medicare utilization to develop the fully implemented RVUs that appear in that final rule. Because these data were unavailable to us for the proposed rule, the first time we could act on this public comment was in the final rule. We have continued our policy of using the latest utilization data to develop each successive year's fully implemented practice expense RVUs during each year of the transition (see 65 FR 65436, published on November 1, 2000, and 66 FR 55322, published on November 1, 2001).

While substituting the latest year's utilization data into the practice expense methodology generally made little difference on total Medicare payments per specialty, it had a larger impact on services that have values affected by the zero physician work pool. The practice expense values for the technical component and other services included in the zero physician work pool declined 4 percent in 2002 as a result of using the most recent Medicare utilization data. Since the technical component is used to derive the global practice expense RVUs for professional and technical component services, there was also a reduction in the practice expense RVU for the global service.

The specialties that provide many of the services that are included in the zero physician work pool have expressed concern about the impact of the most recent data on utilization on values for their services. They recently suggested that we use combined utilization data from 1997 to 2000 to determine the practice expense values. Alternatively, these commenters suggested using either the 1997 or 1999 utilization as a “base year” until an alternative to the zero physician work pool can be developed. These commenters further indicated that, once an option is chosen, we should not use more recent utilization data until comprehensive reform of the zero physician work methodology is adopted.

We believe the suggestion of using multiple years of utilization data in the practice expense methodology has merit. Using multiple years of data has the potential to minimize the effect of year to year case mix changes on practice expense RVUs and improves the stability of our payment systems. We are proposing to develop the practice expense RVUs using Medicare utilization data from 1997-2000. More information on the impact of this proposal can be found in the regulatory impact statement of this proposed rule.

We also agree with the suggestion that the utilization data not change annually until the zero physician work pool is eliminated. In fact, we are reconsidering whether to continue the practice of using the most recent utilization to develop each successive year's practice expense RVUs. As we have indicated elsewhere in this and earlier rules, we are continuing the refinement process beyond the 1998-2002 transition period mandated by the BBA. Once the refinement process is complete, we believe that the physician community has a reasonable expectation that the practice expense RVUs will not change from year to year unless further refinement is undertaken. Once the initial refinement of practice expense RVUs is complete, we expect to make additional refinements at least every 5 years as provided for in section 1848(c)(2)(B) of the Act. As the refinement process continues, there have been fewer widespread changes to Medicare payments and there has been increased year-to-year consistency in the practice expense RVUs. We believe this stability would improve if we incorpoated the most recent utilization data into the practice expense methodology only when we undertake substantial refinement as part of a 5-year review. For this reason, we are proposing to use the 1997-2000 utilization data to develop the CY 2003 practice expense RVUs and not further update the utilization data to incorporate the 2001 utilization data in this year's final rule. Further, we are proposing to continue using the 1997-2000 utilization data in the practice expense methodology until we undertake the 5-year review of practice expense RVUs. We invite comments on these issues.

As part of our resource-based practice expense methodology, we make a distinction between the practice expense RVUs for the non-facility and the facility setting.

This distinction is needed because of the higher resource costs to the physician in the non-facility setting when the practitioner typically bears the cost of the resources associated with the service. In addition, the distinction ensures that we do not make a duplicate payment for any of the practice expenses incurred in performing a service for a Medicare beneficiary. When the beneficiary is a facility patient, we pay the facility for the clinical staff, supplies, and equipment needed to care for the patient. A generally lower facility practice expense rate is paid to the practitioner. Currently, we have designated only hospitals, skilled nursing facilities (SNFs), and community mental health centers (CMHCs) as facilities for purposes of calculating practice expense. An ambulatory surgical center (ASC) is designated as a facility if it is the place of service for a procedure on the ASC list. All other places of service are currently considered non-facility.

Several new places of service are now in use for which we need to assign a site-of-service designation. Also, we are proposing revisions to the site-of-service designation for several existing places of service. We are proposing to assign a facility site of service where a facility or other payment will be made, in addition to the physician fee schedule payment to the practitioner, to reflect the practice expenses incurred in providing a service to a Medicare patient. We are proposing to designate all other places of service as non-facilities.

The following is a list of the new places of service, along with their place of service numerical codes and their proposed site of service designations using the above criteria:

We are proposing that this be considered a nonfacility setting.

We are proposing that this be considered a nonfacility setting.

We are proposing that this be considered a facility setting.

We are proposing that this be considered a nonfacility setting.

We are proposing that this be considered a facility setting.

We are proposing that this be considered a nonfacility setting.

If a mobile unit provides a service to a facility patient, the appropriate place-of-service code for the facility should be used. For instance, if a portable X-ray service is provided to a patient in a Part A skilled nursing facility stay, the place of service is 31, Skilled Nursing Facility. No payment is made under Part B for the technical component of a diagnostic test, portable x-ray transportation or portable x-ray set up. Payment is made to the SNF for Part A services and includes payment for diagnostic services that may be needed by the patient. This policy is consistent with recommendations made by the Inspector General in a recent report, Review of Improper Payments Made by Medicare Part B for Covered Services under the Part A Skilled Nursing Facility Prospective Payment System (A-01-00-00538).

We are proposing that this be considered a nonfacility setting.

We are proposing changes in site of service to the following current designations:

Currently this is designated as a nonfacility. We are proposing that this be considered a facility setting.

Currently codes 41 and 42 are designated as nonfacility. We would propose to designate them as facilities because we make payments for ambulance services using the ambulance fee schedule that covers the direct practice expense.

Currently, this is designated as a nonfacility. We are proposing that this be considered a facility setting.

Currently, this is designated as a nonfacility. We are proposing that this be considered a facility setting.

In the chart below is a complete list of all the existing place-of-service codes along with the appropriate site-of-service designation and the descriptor for each. These codes are used on all professional claims to specify the entity where services are furnished.

Medical and surgical services paid under the physician fee schedule have three separate relative value components, a work RVU, a practice expense RVU and a malpractice RVU. Physician anesthesia services are paid under the physician fee schedule, but the payment method is different than the payment method for physician medical and surgical services. Payment for anesthesia services is based on the sum of base units and anesthesia time units multiplied by an anesthesia CF that is different from the physician fee schedule CF for medical and surgical services.

The law requires that we review RVUs no less than every 5 years. The first 5-year review of work RVUs was completed and the revised work RVUs were implemented in 1997. The second 5-year review (with the exception of anesthesia services) was completed and the revised work RVUs implemented in CY 2002.

In the first 5-year review of work RVUs, we accepted the American Medical Association's (AMA's) Relative Value Update Committee's (RUC's) recommendation that the work of anesthesia services was undervalued by approximately 23 percent. Since anesthesia services do not have individual work RVUs per code, the adjustment in anesthesia work was made to the anesthesia CF and not to the anesthesia codes themselves. This resulted in a 16-percent increase in the anesthesia CF. Budget neutrality was maintained by making an adjustment to the general physician fee schedule.

For the second 5-year review, the American Society of Anesthesiologists (ASA) submitted comments to us contending that the work of anesthesia services is still undervalued by almost 31 percent. The Society subsequently reduced this to a request for a 26-percent increase in work based on additional discussions with the RUC.

We can impute an anesthesia work value from the current allowed charge for an anesthesia service. This work value can be compared to the work value for anesthesia services that is derived from a building block approach. Under the building block approach, uniform individual components of the anesthesia service are identified and the work value of each component is estimated on the basis of a comparable physician medical or surgical service. The ASA derived a work value for an anesthesia code by dividing the anesthesia service into five uniform components and compared the work of each component to a comparable medical or surgical service. The five components are—preoperative evaluation, equipment and supply preparation, induction period, postinduction period, and postoperative care and visits. Using this method, the ASA proposed work values for 19 high volume anesthesia codes. The 19 codes represent a reasonable variety of surgical procedure types, including general surgery, vascular surgery, neurosurgery, urology, orthopedics, cardiac surgery, and ophthalmology. The base units of the 19 anesthesia codes reviewed range from three to twenty units.

During this second 5-year review of work, four RUC workgroups have reviewed the ASA comments and received supplemental information through presentations from the ASA. Most of these workgroups have expressed concerns about some of the intensity values that ASA assigned to the individual anesthesia components, most notably, the induction and postinduction time periods. Each of these workgroups expressed serious concern about extrapolating the imputed work undervaluation from the 19 survey codes to all anesthesia service codes, even though these 19 codes account for more than 40 percent of all anesthesia associated with surgical services.

Despite the efforts of its workgroups, the RUC furnished no recommendation to us on whether the work of anesthesia services is over-or undervalued. In the November 1, 2001 physician fee schedule final rule, we stated that:

The RUC has informed us that it will continue to look at anesthesia work beginning at its first meeting in CY 2002. We will review the RUC recommendation and address anesthesia work in next year's proposed physician fee schedule rule.

The RUC recently presented us with the analysis and findings of its April 2002 anesthesia workgroup. Despite its detailed analysis and laborious discussions of this issue, the RUC concluded that it was unable to make a recommendation regarding modification to the physician work valuation of anesthesia codes. Specifically, the RUC indicated the following:

“The RUC, having carefully considered the information presented, and having a reasonable level of confidence in the data which was presented and developed by the RUC, is unable to make a recommendation to CMS regarding modification to the physician work valuation of anesthesia codes.”

At the April 2002 meeting, the RUC anesthesia workgroup reviewed the postinduction intensity values for the 19 anesthesia codes. The group also reviewed each anesthesia code, the benchmark surgical code, and the five codes mapped to that anesthesia code that accounted for the largest percentage of total volume. The group considered the extent to which the anesthesia work of the benchmark surgical code is representative of other surgical codes that would be covered by the anesthesia code.

We will review the information forwarded by the RUC and all comments we receive during the comment period to determine if an appropriate adjustment can be made to anesthesia work. We would note that any such adjustment would also require an adjustment to the conversion factor for all physicians' services, as required by section 1848(c)(2)(B)(ii) of the Act. For example, a 26 percent increase in anesthesia work, an amount which was requested last year, would require a reduction of about 0.4 percent in the conversion factor for all services. We welcome comments on these issues.

As we discuss above, payment for anesthesia services is based on the sum of an anesthesia code-specific base unit value plus anesthesia time units multiplied by an anesthesia CF. If the physician is involved in multiple anesthesia services for the same patient during the same operative session, payment is based on the base unit assigned to the anesthesia service having the highest base unit value and anesthesia time that encompasses the multiple services. This policy was adopted at the start of the physician fee schedule in 1992 and is incorporated in § 414.48(g).

Claims processing manuals instruct the carrier on the method for handling anesthesia associated with multiple or bilateral surgical procedures. Under Medicare Carrier Manual (MCM) 4830 D, the carrier instructs the physician to report the anesthesia procedure with the highest base unit value with the multiple procedures modifier, “51”, and to report total time across all surgical procedures. Thus, the carrier is recognizing payment for one anesthesia code, despite the billing of multiple surgical codes by a surgeon.

In 2001 and 2002, the CPT has added new anesthesia codes, some of which are add-on codes. The objective is that the add-on code would be billed with a primary code and the base unit of each code would be allowed.

In the burn area, CPT code 01953 (1 base unit) is used in conjunction with CPT code 01952 (5 base units). In the obstetrical area, CPT code 01968 (2 base units) is used in conjunction with CPT code 01967 (5 base units) and CPT code 01969 (5 base units) is used in conjunction with CPT code 01967 (5 base units).

The application of the multiple anesthesia service policy means that the base units of the add-on codes would never be recognized. Only the base units of the primary code would be allowed. We believe that anesthesia add-on codes should be priced differently than other multiple anesthesia codes. As a result, we are proposing to revise the regulations in § 414.46(g) to include an exception to the usual multiple anesthesia services policy for add-on codes.

In its March 2002 Report to Congress, MedPAC recommended that “The Secretary should revise the productivity adjustment for physicians' services and make it a multifactor instead of labor-only adjustment.” In this section, we review the history of the Medicare Economic Index (MEI) productivity adjustment, describe the current MEI productivity adjustment, and identify and evaluate possible alternative MEI productivity adjustments based on the individual contributions we solicited from experts on this topic. We conclude by proposing that the MEI productivity adjustment be changed to reflect an economy-wide multifactor productivity adjustment.

The MEI is based on the fourth sentence of section 1842(b)(3) of the Act that states that prevailing charge levels beginning after June 30, 1973 may not exceed the level from the previous year except to the extent that the Secretary finds, on the basis of appropriate economic index data, that such higher level is justified by year-to-year economic changes. S. Rept. No. 92-1230 at 191 (1972) provides slightly more detail on that index, stating that:

Based on this legislative intent, in 1975, we determined that the MEI would be based on a broad wage measure reflecting overall earnings growth, rather than direct inclusion of physicians' net income. We used average weekly earnings of nonagricultural production (nonsupervisory) workers, net of worker's productivity, as the wage proxy in the initial MEI. We included the productivity adjustment because it avoided double counting of gains in earnings resulting from growth in productivity and produced a MEI that approximated an economy-wide output price index like the Consumer Price Index (CPI). The productivity adjustment we used was the annual change in economy-wide private nonfarm business labor productivity, applied only to the physicians' earnings portion of the MEI (then 60 percent).

As noted, the productivity adjustment in the MEI serves to avoid the double counting of productivity gains. Absent the adjustment, productivity gains from producing additional outputs (procedures) with a given amount of inputs would be included in both the earnings component of the MEI (reflecting growth in overall economy-wide productivity) and in the additional procedures that are billed (reflecting physicians' own productivity gains). Therefore, general economic labor productivity growth is removed from the labor portion of the price update.

The basic structure of the MEI remained relatively unchanged from its effective date (July 1, 1975) to 1992, although its weights were updated periodically and a component was added for professional liability insurance. Section 9331 of the Omnibus Budget Reconciliation Act of 1986 (Pub. L. 99-509) (OBRA) mandated a study of the MEI and a notice and opportunity for public comment before revision of the methodology for calculating the MEI. Based on this requirement, we held a workshop with experts on the MEI in March 1987 to discuss topics ranging from the specific type of index to use (Laspeyres versus Paasche) to revising the method of reflecting productivity changes. Participants in the meeting included the Federal government, the Physician Payment Review Commission (PPRC), the Congressional Budget Office, the American Medical Association (AMA), and several consulting firms. The meeting participants concluded that a productivity adjustment was appropriate and that an acceptable measure of physician-specific productivity did not exist. Many alternative approaches were discussed, including the use of a policy-based “target” measure and several existing economic productivity measures.

Using recommendations from the meeting participants, we revised the MEI and the productivity adjustment with the implementation of the physician fee schedule as discussed in the November 1992 final rule (57 FR 55896). While we retained an adjustment for economy-wide labor productivity, it was applied to all of the direct labor categories of the MEI (70.448 percent), not just physicians' earnings, and was based on the 10-year moving average percent change (instead of annual percent changes). This form of the index has been used since that time, and was most recently discussed in the November 1998 final rule (63 FR 58845) when the MEI weights were rebased to a 1996 base year.

The Balanced Budget Act replaced the Medicare Volume Performance Standard (MVPS) with a Sustainable Growth Rate (SGR). Section 1848(f) of the Act specifies the formula for establishing yearly SGR target for physicians' services under Medicare. The use of SGR targets is intended to control the actual growth in aggregate Medicare expenditures for physicians' services. The SGR targets are not limits on expenditures. Payments for services are not withheld if the SGR target is exceeded by actual expenditures. Rather, the appropriate fee schedule update, as specified in section 1848(d)(3) of the Act, is adjusted to reflect the success or failure in meeting the SGR target. If expenditures are less than the target, the update is increased. If expenditures exceed the target, the update is reduced. Specifically, expenditures are allowed to increase by fee-for-service Medicare enrollment growth, physician fee increases, increases in real per capita Gross Domestic Product (GDP), and changes in laws or regulations. Consequently, the statute links allowable increases in the volume of services resulting from physician productivity gains—together with volume and intensity increases due to technology and other factors—to the real per capita GDP.

When the SGR was enacted, the Congress specified continued use of the MEI. By 1997, this index, including its productivity adjustment, had been used in updating Medicare payments to physicians for over twenty years. We did not propose any changes to the productivity adjustment used in the MEI because its continued use was consistent with the newly mandated SGR. If we did not make the adjustment in the MEI, general economic productivity gains would be reflected in two of the SGR factors, the MEI and real per-capita GDP (which reflects real GDP per hour worked, or labor productivity, and hours worked per person). We believe it is reasonable to remove the effect of general economic productivity from one of these factors (the MEI) to avoid double counting.

The current MEI productivity adjustment is based on the 10-year moving average percent change in private nonfarm business (referred to hereafter as “economy-wide”) labor productivity, published by the Bureau of Labor Statistics (BLS) on a quarterly basis. A 10-year moving average is used to limit the impact of cyclical fluctuations in productivity. The productivity adjustment is applied only to the direct labor portions of the MEI (currently estimated at 71.272 percent). Therefore, the MEI is not reduced by the full change in labor productivity, but instead by only a portion of the change.

In addition, the most recently available historical data are used for the update for the upcoming calendar year (for example, data available through the second quarter of CY 2001 was used for the CY 2002 update).

Under this method, the current estimate of the existing MEI for the CY 2003 fee schedule update would be 2.3 percent. The 10-year moving average percent change in economy-wide labor productivity for the CY 2003 update is estimated to be 2.1 percent. However, since this adjustment is applied only to the direct labor portion of the MEI, the actual adjustment would be 1.5 percent. By comparison, the most recent forecast by DRI-WEFA, a Global Insight Company, of the CPI for all items for this same period is 1.6 percent.

As noted previously, since its original development, the MEI productivity adjustment has been based on economy-wide productivity changes. This practice arose from the fact that the physicians' compensation portion of the MEI is proxied to grow at the same rate as general earnings in the overall economy, which reflect growth in overall economy-wide productivity. Removing labor productivity growth reflected in general earnings from the labor portion of the MEI produces an index that is consistent with other economy-wide output price indexes, like the CPI. Although some commenters have argued that use of a physician-specific productivity measure would be more appropriate, no such published measure existed at the time of the MEI's development; nor does one exist today.

We conducted a number of research activities to evaluate whether the current productivity adjustment is still the most appropriate adjustment to use in the MEI. First, we evaluated the currently available productivity estimates that are produced by the BLS to develop a better understanding of the strengths and weaknesses of these measures. We also reviewed the theoretical foundation of the MEI to understand how labor and multifactor productivity relate to the current physician payment system. Then we studied the limited publicly available data to begin to develop preliminary estimates of trends in physician-specific productivity to better understand the current market conditions facing physicians. Finally, we solicited the individual contributions of academic and other professional economic experts on prices and productivity. They included experts from MedPAC, AMA, OMB, Dr. Uwe Reinhardt from Princeton University, Dr. Joe Newhouse from Harvard University, Dr. Ernst Berndt from MIT, and Dr. Joel Popkin from Joel Popkin and Company (former Assistant Commissioner of Prices at BLS). Based on the information we gathered during these research efforts, we evaluated six possible options for a productivity adjustment to the MEI. Our findings on each of the options we investigated are summarized below:

• Option 1—Using a physician-specific productivity adjustment.

This option would entail using an estimate of physician-specific productivity to adjust the MEI. This option may have some theoretical attractiveness, but there are major problems obtaining accurate measures of physician-specific productivity. First, no published measure of physician-specific productivity is available. The Federal agency that produces the official government statistics on productivity, BLS, does not calculate or publish productivity measures for any health sector. Nor are there alternative measures of physician-specific productivity that incorporate the BLS methodology of measuring productivity and that would meet the BLS standard of publication. Second, it is not clear that using physician-specific productivity within the current structure of the MEI would be appropriate. Because we believe the MEI appropriately uses an economy-wide wage measure as the proxy for physician wages, using physician specific productivity could overstate or understate the appropriate wage increase in the MEI.

We do believe, however, that it is important to understand the rate of change in physician-specific productivity. Toward this end, we have performed our own preliminary analysis of physician-specific productivity, using the limited publicly available data on physician outputs and inputs. Our analysis attempted to simulate the methodology the BLS would use to measure productivity. While this information cannot be interpreted as an official measure of productivity, we do believe it is a rough indication of the current market conditions facing physicians. We used this information to help form our determination of the most appropriate productivity adjustment to incorporate in the MEI, fully recognizing its preliminary nature and other limitations. The results of our preliminary analysis suggest that long-run physician-specific productivity growth is currently at approximately the same level as economy-wide multifactor productivity growth. Prior to the recent period, however, our preliminary estimates suggested that physician productivity gains were generally significantly greater than general economy-wide multifactor productivity gains.

As we have emphasized, our rough estimates are inadequate for establishing a formal basis for the productivity adjustment to the MEI. Nor is the underlying economic theory sufficiently compelling, at this time, to adopt a physician-specific productivity measure, even if a suitable one were available. We conclude, however, that economy-wide multifactor productivity growth appears to be roughly comparable to current physician-specific productivity growth.

• Option 2—Retaining the current productivity adjustment.

We investigated retaining the current productivity adjustment, that is, applying the 10-year moving average percent change in economy-wide labor productivity to the labor portion of the MEI. We have applied economy-wide labor productivity to a portion of the index in some form since the inception of the MEI in 1975. This current form has been used since the last major revision to the index in 1992 and was developed from the contributions of the 1987 expert panel. That panel concluded that using labor productivity applied to the labor portion of the index was a technically sound way to account for productivity in the physician update. This method makes optimal use of the available data since labor productivity data were, and are, available on a more timely basis than economy-wide multifactor productivity. By applying this measure to the labor portion of the index, the mix of physician-specific labor and nonlabor inputs is reflected. Also, the use of a 10-year moving average percentage change reduces the volatility of annual labor productivity changes.

Our research, however, has indicated that using multifactor productivity applied to the entire index is superior to using an economy-wide labor productivity measure applied only to the labor portion of the index. The experts with whom we consulted believed it was more appropriate to reflect the explicit contribution to output from all inputs. The current measure explicitly reflects the changes in economy-wide labor inputs but does not reflect the actual change in nonlabor inputs. Instead, it implicitly assumes that nonlabor inputs would grow at the rate necessary to produce an economy-wide multifactor measure that is equivalent to the current MEI productivity adjustment. That implicit assumption is less precise than a direct, explicit calculation.

In addition, while the implicit approach produced an MEI productivity adjustment in most years that was reasonably consistent with overall multifactor productivity growth, it now appears less consistent with the actual change in nonlabor inputs in the economy. In recent years, economy-wide labor productivity has grown very rapidly. This acceleration is partly the result of major investments in computers (a nonlabor input) that have helped create a more productive work force. Also, the Bureau of Economic Analysis (BEA) has adopted methodological changes in accounting for computer software purchases in measuring GDP. These changes have significantly increased the measured historical growth rates in real GDP and labor productivity. As a result of these developments, the MEI productivity adjustment based on labor productivity applied only to the labor portion of the MEI has increased very rapidly. Since the multifactor definition is an explicit calculation of the change in economic output relative to the change in both labor and nonlabor inputs, it better reflects the trend changes.

Finally, as noted previously, our preliminary estimates of physician-specific productivity suggest a current growth pattern that is similar to growth in multifactor productivity in the economy overall. In consideration of the economic theory underlying productivity measurement, especially in view of the recent developments in labor versus nonlabor economic input growth trends, we concluded that using a multifactor productivity adjustment is superior to the current methodology for adjustment for productivity in the MEI.

• Option 3—Changing to using economy-wide multifactor productivity.

One option for adjusting for productivity gains in the MEI would be to continue to use an economy-wide productivity measure, but to use multifactor productivity applied to the entire index, instead of labor productivity applied to the labor portion of the MEI. As noted previously, this approach was recommended by MedPAC in its March 2002 Report to the Congress. This option would better satisfy the theoretical requirements of an output price, in this case the MEI, by explicitly reflecting the productivity gains from all inputs. In addition, the use of economy-wide multifactor productivity would still be consistent with the MEI's use of economy-wide wages as a proxy for physician earnings. While annual multifactor productivity can fluctuate considerably, though usually less than labor productivity, using a moving-average would produce a relatively stable and predictable adjustment.

Each expert with whom we consulted believed that using a multifactor productivity measure was theoretically superior to the existing method because it reflected the actual changes in nonlabor inputs instead of reflecting an implicit assumption. They also believed that the lack of timely data on multifactor productivity was not as important as would have appeared initially. Instead, the experts believed it was more appropriate that the adjustment be based on a long-run average that was stable and predictable rather than on annual changes in productivity. Thus, if a long-run average were used, the increased lag time associated with the availability of published data on multifactor productivity would become less significant. Finally, one expert believed that changing to economy-wide multifactor productivity applied to the entire MEI would make it easier to understand the magnitude of the productivity adjustment.

Use of multifactor productivity to adjust the MEI poses two concerns. First, multifactor productivity is much harder to measure than labor productivity. Economic inputs other than labor hours can be very difficult to identify and calculate properly. The experts at BLS, however, have adequately overcome these difficulties, and we are satisfied that their official published measurements are sound for the purpose at hand. Moreover, use of a 10-year moving average increase helps to mitigate any remaining measurement variation from year to year.

The second concern relates to the timeliness of the data. BLS publishes multifactor productivity levels and changes only annually (as opposed to the quarterly release of labor productivity data) and with an extended time lag (about 1 1/2 years). These timeframes arise unavoidably from the difficulties of measurement mentioned above, but imply that the timeframe of data used to adjust the MEI would not match that of the historical data on wages and prices underlying the MEI. For the CY 2003 physician payment update, for example, we would use data on wages and prices through the second quarter of CY 2002, but would have to use multifactor productivity data only through CY 2000. Although the misalignment of data periods is a concern, we believe it is a reasonable trade-off in view of the improvement offered by the explicit measurement of nonlabor inputs. Also, since use of a 10-year moving average is intended to reduce fluctuations and provide a more stable level of the productivity adjustment, availability of the most recent data is of less importance.

The 10-year moving average percent change in economy-wide multifactor productivity that would be used for the CY 2003 update (historical data through CY 2000) is currently estimated at 0.8 percent. Our preliminary internal analysis of physician-specific productivity gains suggests that these economy-wide multifactor measures are somewhat consistent with those trends. Thus, using economy-wide multifactor productivity for MEI productivity adjustment theoretically would be superior to using labor productivity growth applied to the labor portion of the MEI. In addition, the use of a 10-year moving average would help alleviate the lag in the availability of the data. Lastly, the current 10-year moving average growth in economy-wide multifactor productivity appears to be within the range we have estimated for physician-specific multifactor productivity. One possible weakness of using economy-wide multifactor productivity is that it does not reflect physician-specific measures, whereas the existing methodology reflects the distribution of labor and nonlabor inputs used in the production of physician services. In practice, however, the balance between these factors of production is not substantially different for physician practices versus the overall economy.

• Option 4—Changing to using economy-wide multifactor productivity with physician-specific input weights.

Another option we explored was using economy-wide labor and capital productivity measures (which, when weighted together, produce multifactor productivity), but with physician-specific input weights. This method would better reflect the proportion of labor and capital inputs used by physicians, yet still reflect the explicit contribution to productivity of labor and nonlabor inputs. The experts with whom we discussed this option thought it was theoretically consistent with a measure of multifactor productivity, even though different productivity measures would be applied to different components of the MEI.

As noted above, the labor and capital shares for the overall economy do not appear to vary enough from the physician-specific shares in the MEI to result in a significantly different measure. A weakness of this method is that the BLS capital productivity series is not widely used or cited; therefore, we are unsure of the accuracy and reliability of this measure. This method also adds another layer of complexity to the formula, however, making it more difficult to understand the adjustment. We would prefer that any method we choose be straightforward so that everyone can readily understand the adjustment. Overall, we believe that this method does not provide enough of a technical improvement to justify the added complexity that would be required to implement it.

• Option 5—Adjusting productivity using a “Policy Standard”.

In its March 2002 Report to the Congress, MedPAC suggested establishing a policy target for the productivity adjustment. Under this methodology, the level of the policy target would be based on the productivity gains that we believe physicians could attain. This level would be set through policy and would likely be based on a long-run average of either economy-wide labor or multifactor productivity (but could reflect other, possibly judgmental, factors). Generally, the level of the policy standard would remain constant for several years; periodically, the policy target would be reviewed, and possibly adjusted.

Some of the experts we consulted believed that a policy target would lessen the volatility of the adjustment since the target would not be changed often. Conversely, others noted the large, abrupt changes that could result if actual economic performance deviated from the policy standard requiring subsequent adjustments to the standard. Some believed that this method adjusts for the problem of precisely measuring productivity. If we used a policy standard we could avoid having to develop an exact measure. Using a policy target, however, may appear arbitrary without a theoretical basis to support its use.

The policy target recommended by MedPAC was 0.5 percentage points per year. Its justification for this number was the fact that the long-run average of economy-wide multifactor productivity was close to 0.5 percent (the most recent 10-year average is now 0.8 percent). We do not believe this is a preferred option for adjusting the MEI for productivity improvements. Our preference is to use a long-term data-based approach that will produce results that are not inconsistent with a policy standard and that will automatically reflect changes in actual economic performance over time, and not through abrupt periodic large adjustments. Thus, we conclude that a policy target does not provide an improvement over any of the data-based methodologies.

• Option 6—Eliminate Productivity Adjustment from the MEI.

Questions are raised occasionally as to the possibility of eliminating the productivity adjustment from the MEI. We did not consider this to be a viable option. Our research concluded that adjusting for productivity in the MEI is necessary to have a technically correct measure of an output price increase, free of double-counting the impact of productivity. Every expert with whom we consulted agreed that a productivity adjustment was appropriate. They believed that the important question is which adjustment is the most appropriate. Therefore, we conclude, again, that it is not acceptable for the productivity adjustment to be removed from the MEI.

MedPAC, in its March 2002 Report to the Congress, recommended the use of a forecasted MEI value, rather than the current historical increase. However, implementation of this option raises several legal as well as practical issues. The 1972 Senate Finance Committee report language reflects Congress' intent that the MEI should “follow rather than lead” overall inflation. Because of this, updates to the physician fee schedule have always been based on historical, rather than forecasted, MEI data. In this way, increases in the MEI do not lead the current measures of inflation but follow them based on historical trends. Furthermore, at the time of implementation of the SGR system, the Congress specified that the SGR system should use the MEI that existed at that time, which was based on historical data measures. The law did not recommend or specify a change in the MEI methodology; the assumption is that the Congress was satisfied that the MEI was functioning as designed.

If we were to change to a forecasted MEI and productivity adjustment, there are also several practical issues that would need to be addressed. One is that changing from a historical-based MEI to a projected MEI would cause transitional problems because there would be a period of data that would not be accounted for in the year of implementation. For example, the CY 2002 MEI update was based on historical data through the second quarter of 2001. If we were to use a forecasted MEI in the update for CY 2003, the changes between the second quarter of 2001 and the first quarter of 2003 would not be accounted for in the update. Finally, changing to a forecasted MEI and productivity adjustment raises additional questions about correcting for forecast errors. Based on these problems, we will continue to use historical data to make updates under the physician fee schedule.

Based on the research we conducted on this issue, we are proposing to change the methodology for adjusting for productivity in the MEI. We propose that the MEI used for the CY 2003 physician payment update reflect changes in the 10-year moving average of private nonfarm business (economy-wide) multifactor productivity applied to the entire index. The current method accounts for productivity by adjusting the labor portion of the MEI by the 10-year moving average change in private nonfarm business (economy-wide) labor productivity.

We propose to make this change because: (1) It is theoretically more appropriate to explicitly reflect the productivity gains associated with all inputs (both labor and nonlabor); (2) the recent growth rate in economy-wide multifactor productivity appears more consistent with the current market conditions facing physicians; and (3) the MEI still uses economy-wide wage changes as a proxy for physician wage changes. We believe that using a 10-year moving average change in economy-wide multifactor productivity produces a stable and predictable adjustment and is consistent with the moving-average methodology used in the existing MEI. We propose that the adjustment be based on the latest available actual historical economy-wide multifactor productivity data, as measured by BLS. Based on these proposed changes, we currently estimate the MEI to increase 3.0 percent for CY 2003. This is the result of a 3.8-percent increase in the price portion of the MEI, adjusted downward by a 0.8-percent increase in the 10-year moving average change in economy-wide multifactor productivity. Table 1 shows the detailed cost categories of the proposed MEI update for CY 2003. Since the current estimate of the MEI increase for CY 2003 is based on incomplete historical data, it may change slightly before we announce the final MEI no later than November 1, 2002.