• The Lewin Group also recommended that we revise edits and trims to the SMS survey data, both practice expenses and hours, to exclude data that fall outside set acceptable ranges (for example, three standard deviations from the geometric mean). We asked the AMA about their reaction to The Lewin Group's recommendation and the AMA replied:

For this reason, we are not taking action in response to The Lewin Group's recommendation at this time.

• In addition, The Lewin Group recommended that we account for item non-response to questions related to practice expenses and patient care hours. We asked the AMA for their reaction to this recommendation as well. The AMA replied that they would need more information and added that there is no evidence that a pattern of non-response bias exists for practice expense, although it is a possibility. We are considering whether to study this issue further but, at this time, are not making any adjustments in response to this recommendation.

The report also makes suggestions on changes to the survey instrument used to collect practice expense data from practitioners. Though the original SMS survey does collect some information on practice expenses, it was not designed as a vehicle to calculate a specialty-specific practice expense per hour. We, and the contractor, have held several meetings with the AMA's SMS staff to discuss revisions to the survey that would help make our calculations more precise.

We understand that the AMA is currently piloting a new practice-level survey designed to address some of the limitations of the SMS. If the pilot of the survey is successful, we earlier understood that the AMA plans were to conduct the practice survey initially in CY 2000 and, in alternate years thereafter, the practice expense survey and the SMS survey. The AMA has recently indicated that its plans about the future of the SMS and collection of practice level survey data are unclear at this time. While the AMA has not made a final decision at this time about whether the practice level survey will be done, they have indicated concern to us about low response rates from the pilot test. Nevertheless, we are proceeding to make recommendations to the AMA regarding collection of practice expense data through the practice level survey. We will continue our discussion with the AMA regarding its plans for future practice expense data collection following completion of the practice level survey. And, as we stated earlier, we believe these recommendations will be useful in the design of the practice level survey or any other survey of practice expenses used in developing RVUs for practice expenses.

The use of this practice level survey, as it is currently contemplated, responds to several of our contractor's recommendations. For example, it would address the recommendation that information be collected on each physician's percent share of practice expense and hours within the practice by collecting information at the total practice, rather than the individual physician owner level. The practice level survey also currently contains, as requested, questions on the number of hours the physician's office is open in a typical week and on the salaries for the mid-level practitioners used by the practice (that is, physician assistants, nurse practitioners, clinical nurse specialists, nurse mid-wives, certified registered nurse anesthetists, and physical and occupational therapists).

We are also suggesting additional changes in the survey questions or directions, generally reflecting our contractor's recommendations. We believe that the following changes would give more precise and reliable data on which to base our practice expense calculations:

• Emphasize the benefit of involving the practice manager or accountant in the completion of the survey and the need to link the practice expense data to the practice's tax information whenever possible.

• Include a question concerning how many patient care hours are spent on uncompensated care, that is, care that the law requires one to provide, but for which one is not compensated. This would not include charity care that is voluntarily provided.

• Add a question concerning the amount or percentage of revenue generated by mid-level practitioners.

• Add a question concerning the amount or percentage of supply costs that relates to separately billable supplies (for example, drugs, casting supplies, and laboratory supplies).

• In addition, we are recommending that the survey include more specific questions on patient care hours and that separately billed mid-level practitioner hours be included.

The Lewin Group also recommended that the survey include questions about a typical week, rather than the most recent week. We are not adopting this suggestion because we believe that questions about the most recent week are likely to yield more concrete, accurate answers, whereas questions about a typical week are more likely based on estimates. As we have already stated, the AMA will no longer be collecting data through the SMS and the AMA has also expressed concern about low response rates from the pilot of the practice level survey. At this time, we are unclear as to the AMA's plans with regard to future practice expense data collection efforts.

As we indicated earlier, we are currently proposing to use data from the 1998 SMS in developing the 2001 practice expense relative value units. Furthermore, data from the 1999 SMS will become available later this year. In addition, section 1848(c)(2)(B) of the Act requires that not less often than every 5 years, we review and make adjustments to RVUs. Thus, by law we are required to review and make adjustments to the practice expense RVUs no later than 2007. Regardless of whether the AMA continues to collect data on practice expenses, we will be developing plans for making refinements to practice expense RVUs beyond 2002.

We welcome comments on long-term strategies for refining the practice expense RVUs and any suggestions for how to collect practice expense data in the event it is no longer collected by the AMA. We will consider these comments and any further decisions by the AMA with regard to its practice expense data collection efforts in developing our refinement strategy beyond 2002.

We have received many comments from specialty societies concerning our calculation of direct patient care hours. This is a major issue because the patient care hours are one half of the ratio used to determine the practice expense per hour for each specialty. (The practice expenses of practitioners in a specialty are divided by the direct patient care hours in order to calculate the practice expense per hour). If the reported hours do not reflect the actual average billable hours for a specialty, the practice expense per hour will be over-or understated.

Several commenters representing surgical specialty societies have raised concern that the hours computed for their specialties have been overstated. This may be a result of SMS survey respondents including non-billable hours (such as stand-by time) when asked how many hours they worked each week. If this is the case, this would decrease the practice expense per hour for these specialties. In addition, commenters representing emergency room physicians raised the issue that the hours spent on uncompensated care were probably also included in the survey responses to the detriment of this specialty.

We agree with the commenters that there is a need to increase the level of confidence in the direct patient care hour data. We are already taking steps to improve the future accuracy of these data. As mentioned above, we are recommending that the future survey questions be worded more precisely so that only the appropriate practitioner hours are included. In addition, we have asked our contractor to give priority to recommendations on steps we can take to improve the accuracy of the patient care hours.

As a first step in accomplishing this, The Lewin Group issued their second draft report on December 6, 1999, entitled “Validating Patient Care Hours Used in HCFA's Practice Expense Methodology.” This report explores alternative methods that we might use to validate the time data collected by the SMS survey. The validation techniques attempt to achieve two goals: (1) Identifying inaccurate existing data and (2) identifying inconsistencies in new data to be derived from future survey efforts.

The Lewin Group developed the following four validation techniques to analyze the SMS data used in computing the specialty-specific patient care hours:

• Method 1: Compare the patient care hour data reported at the beginning of the SMS survey (that asks for the total hours worked in a week) to responses from the detailed questions on patient care hours appearing later in the SMS survey.

• Method 2: Calculate ratios of SMS time pools to Harvard/RUC time pools by specialty, using Harvard/RUC procedure time data and Medicare claims data.

• Method 3: Compare newly reported SMS data to historical SMS data to identify outliers.

• Method 4: Compare SMS data on annual hours worked with annual hours data reported in the Medical Group Management Association's (MGMA) “Physician Compensation and Production Survey”.

We have placed this report on our homepage under the title “Validating Patient Care Hours.”

We agree with our contractor that no single validation approach exists that can be used to validate both existing and new data on patient care hours with a high level of confidence. However, the approaches described above, when used together, could be effective tools that will help to ensure the accuracy and reliability of existing and future data used in the calculation of practice expense RVUs. These validation efforts would allow us, and the medical community, to be more confident in the use of future data to update practice expense RVUs. Therefore, we extended The Lewin Group's contract so that, among other refinement tasks, the above analyses can be carried out. We are aware that even with the above initiatives, it might not be possible to address all concerns regarding refinement of the patient care hours in the short term. Therefore, we welcome any comments and suggestions as to other steps we could take to verify and improve the accuracy of the specialty-specific patient care hours.

(c) CPEP Data .

The PEAC, a subcommittee of the RUC, held its initial meetings last year and the RUC made recommendations on CPEP inputs for clinical staff times, supplies, and equipment on approximately 65 CPT codes. We discussed our actions with regard to these recommendations in the November 1999 final rule. The PEAC continues to meet to refine the CPEP direct cost inputs, and we anticipate that we will receive additional RUC recommendations in July. We will address these recommendations in this year's physician fee schedule final rule.

In the November 1999 final rule, we deferred action on the RUC recommendations for a few groups of CPT codes on which we had significant questions. We are now proposing to accept the RUC recommendations with the revisions noted below:

We are accepting the total clinical staff time recommended for the in-office setting, but are moving 60 minutes from post to intra-service time for each of the above procedures because the staff time for observation of the patient during recovery from anesthesia belongs in the intra-service period. We are reducing the out of office preservice clinical staff time for CPT codes 52647 and 53852 to 30 minutes to match the RUC recommendation for CPT code 53850 and the time allotted in the office for each service and are making the out-of-office postservice time equal to the in-office postservice time because we believe there is no reason that these times should differ.

The supplies for all three procedures were adjusted to reflect three postoperative visits and to conform with the overall adjustment to supplies made in the November 1999 final rule. For CPT code 52647, we deleted the flexible cystoscope from the equipment because only one scope is required for the procedure. We also deleted the sterilizer because it is not typically used. For CPT code 53850, the RUC recommendations included the inputs for two different scenarios using two different devices. We chose what we believe to be the most typically used device and the inputs that accompany this. For CPT code 53852, we deleted the cystoscopes and sterilizer from the equipment because we believe that they are not typically used.

The RUC had recommended 102 minutes of clinical staff time for CPT code 96408 and 121 minutes for CPT code 96410. In the November 1999 final rule, we solicited comments on these codes to assist us in our review. In response, the American Society of Clinical Oncology provided a breakdown by specific tasks of the above staff times. Included in this breakdown were 20 minutes for pre- and postprocedure education and 15 minutes for three phone calls after each visit.

Because we believe that the times for patient education and phone calls should be averaged over the whole course of chemotherapy treatment, and because there appeared to be some duplication in the pre- and postprocedure education tasks, we reduced both the patient education and phone call times by 5 minutes. Therefore, we are proposing 92 minutes of clinical staff time for CPT code 96408 and 111 minutes for CPT code 96410. For supplies, the specialty society agreed that we should delete the silver nitrate stick and HEPA filters from both procedures and the infusion pump cassette from CPT code 96408.

In the November 1999 final rule, we removed estimates of all clinical staff time allotted to the use of clinical staff in the facility setting from the CPEP data. Commenters have since noted that the clinical staff times reported by some CPEP panels for pre- and postservice times for 0-day global services performed in the office were recorded in the intra-service field in the CPEP database. These times were, therefore, deleted along with the times for the use of clinical staff in the facility setting, unlike the pre- and postservice times for 10 and 90-day global services that were entered into the separate pre and post data fields. The commenters argued that these pre- and postservice staff times for the relevant 0-day global services should be reinstated because these times are for staff in the office before and after the patient is in the facility.

We agree that these data are not comparable to the data we excluded for clinical staff used in the facility setting. We reviewed the “CPEP Recorders' Notes Files” compiled for each CPEP panel by Abt Associates, Inc., the contractor managing the CPEP panels. When the notes indicate that clinical staff estimates were for activities performed in physicians' offices, we are proposing to reinstate the time data for 0-day global services. The fact that we have reinstated these time data does not mean that we necessarily agree that the amount of time assigned is correct. Like all the other raw CPEP data, these time data are subject to refinement and possible revision.

The entire recorders' notes file is available on our website and is entitled “CPEP Recorders' Notes Files.” Addendum C shows a list of the codes for which pre- or postclinical staff time has been added, as well as the times that are now assigned.

In the November 1999 final rule (64 FR 59392), we indicated that casting materials are bundled into the payment for the initial fracture management procedures and that separate billing for the supplies is not allowed. However, commenters noted that our policy has been to allow separate payment for splints, casts, and other devices used for the reduction of all fractures and dislocations under section 1861(s)(5) of the Act. Since we provide separate payment for splints and casting supplies, we are now proposing to remove these types of expenses from practice expense inputs for all applicable fracture management and cast/strapping application procedure codes under the physician fee schedule.

In the November 1999 final rule, we deleted certain casting supplies (fiberglass roll, cast padding, and cast shoe) from the list of supplies for the casting and strapping CPT codes 29000 through 29750. We have identified additional CPT codes for the treatment of fractures/dislocations that have these supplies included in the CPEP data. Since these supplies are currently separately billable, we are proposing to remove the fiberglass roll, cast padding, and cast shoe from the following CPT codes: 23500 through 23680; 24500 through 24685; 25500 through 25695; 26600 through 26785; 27500 through 27566; 27750 through 27848; and 28400 through 28675.

In addition, we are also proposing to remove additional casting and splinting supplies from all the CPT codes referenced above because these supplies are also currently separately billable under section 1861(s)(5) of the Act. The list of supplies is as follows: stockingnet/stockinette; plaster bandage; Denver splint; dome paste bandage; cast sole; elastoplast roll; fiberglass splint; Ace wrap; Kerlix; Webril; Malleable Archbars; and elastics.

We welcome comments on whether these supplies should be deleted from additional procedures outside the code ranges referenced above, and whether we have appropriately identified all the casting supplies in our supply list.

We are currently using the original CPEP definitions for equipment that distinguish between “procedure specific equipment” and “overhead” equipment. The main distinction between the two categories is that procedure specific equipment is used only for a limited number of procedures, while overhead equipment is used over a wide range of services. In terms of actual application, we assume a 50-percent utilization rate for procedure specific equipment, but a 100-percent rate for all overhead equipment. In addition, the methodology assumes that the procedure specific equipment is used only during the intraservice period, while it assumes that the overhead equipment is used for the entire service. We believe this distinction was more important under our original “bottom-up” methodology when the accuracy of the practice expense RVUs was almost totally dependent on the precision of the CPEP inputs. Under our current “top-down” methodology, however, when the CPEP inputs are used only as allocators of the specialty-specific practice expense pools, the distinction has served to hinder the process of refining the CPEP inputs while not leading to a substantive distinction in how we value services.

We are proposing to combine both categories of equipment into a single “equipment” category, assuming an average 50-percent utilization for all equipment. We believe that this will be beneficial to our refinement process for the following reasons:

• The current definition of the two categories of equipment necessitates many subjective decisions. While it might be obvious that an examination table is used for a wide range of services and, therefore, would be overhead equipment, it is somewhat more arbitrary to classify equipment such as cystoscopes or specific x-ray machines as overhead or procedure specific.

• The various CPEP panels were not consistent in their application of the distinction between the two categories. Most of the items that were classified by some of the CPEP panels as overhead equipment were classified by another panel as procedure specific. In addition, equipment that would seem to be very similar was sometimes treated in different ways. For example, an examination table or a stretcher were considered to be overhead, but an electric table or a wheelchair were considered procedure-specific.

• It would simplify the refinement process to have only one category of equipment to consider rather than having to decide for all 7000 codes to which category each piece of equipment belongs.

We are also proposing to delete from the CPEP data equipment that is not used typically with any service, but is on “standby” for many services, or that is used for multiple services at the same time. In either of these cases, it is difficult to allocate the cost of this equipment appropriately to individual CPT codes. Examples of “standby” equipment are crash carts, defibrillators, wheelchairs, and stretchers. Examples of equipment used for multiple procedures at the same time are cabinets, refrigerators, and autoclaves.

Following is the list of equipment that we are proposing to delete at this time from the CPEP inputs of all services: autoclave, wheelchair, refrigerator, film file cabinet, hazard material spill kit, embryo freezer, water system, flammable reagent cabinet, utility freezer, ultra low temperature freezer, acid cabinet, bulk storage refrigerator, abortion clinic security system, abortion clinic security guard, gomco suction machine, doppler, laser printer, lead shielding, defibrillator with cardiac monitor, blood pressure/pulseox monitor, blood pressure monitor, printer, crash cart—no defibrillator, and smoke evacuator.

The following is a list of equipment that we are proposing to delete as “standby” equipment for most codes, but that we believe typically may be used with a designated subset of procedures:

• X-ray view box—four panel (retain when currently in the CPEP data for codes in the range CPT codes 70010 through 79999).

• ECG machine—3 channel (retain when currently in the CPEP data for CPT codes 93000 through 93221).

• Pulse oximeter (retain when currently in the CPEP data for CPT codes 94620, 94621, 94680, 94681 and 94690; 94760 through 94770, 95807 through 95811 and 95819).

• ECG/blood pressure monitor—3 channel (retain when currently in the CPEP data for CPT codes 43200 through 43202 and 43234 through 43239).

• Cardiac monitor (retain when currently in the CPEP data for CPT codes 31615 through 31628).

• ECG-Burdick (except for HCPCS code G0166).

We welcome comments on this proposal and on any additional equipment that should not be considered a direct expense because the cost cannot appropriately be allocated to an individual service. Neither of these proposals to improve the CPEP equipment data have a significant impact on any specialty.

In the November 1999 final rule, we made corrections to the CPEP data for a number of codes that we learned contained errors and anomalies that we could easily correct. Since that time, we have discovered some additional anomalies, and we are proposing to correct them at this time. As we stated in the final rule, though certain revisions may be made now, all practice expense inputs for these codes are still subject to further comment, refinement, and potential PEAC and RUC review and recommendations.

• We have identified several CPT codes that were not costed by the CPEP panels and were not assigned CPEP inputs. We are now crosswalking these services to the CPEP inputs of the most appropriate other service. The CPEP inputs for these codes are subject to refinement. We welcome comments on the crosswalks that we have chosen. The codes and their crosswalks are shown below:

• Because the following either are not performed in the office setting or because we do not have appropriate CPEP inputs for the in-office setting for these services, we are designating the following CPT and HCPCS codes as “N/A” in the office setting: 99183 (Hyperbaric oxygen therapy); 21493 (Treatment of hyoid bone fracture); 21494 (Treatment of hyoid bone fracture with manipulation); 32997 (Total lung lavage); 33968 (Remove aortic assist device); 66830 (Removal of lens lesion); 69990 (Micro-surgery add-on); 92961 (Cardioversion, electric, internal) and we are designating G0167 (Hyperbaric oxygen treatment; no physician required) as carrier priced.

• The TC for CPT code 93660 (Tilt table evaluation) is carrier priced, but we are proposing to price it nationally. Therefore, we are reinstating the original CPEP data.

• We are crosswalking all CPEP inputs for CPT code 44201 (Laparascopy, jejunostomy) from the inputs for CPT code 44200 (Laparoscopy, enterolysis) to reflect that it is a 90-day global service.

• We are adjusting the CPEP inputs for CPT codes 15001 (Skin graft add-on); 15351 (Skin hemograft add-on); and 15401 (Skin heterograft add-on) to reflect that these are ZZZ services.

• CPT code 00103 (Anesthesia for blepharoplasty), which was not costed by the anesthesia CPEP panel, was inadvertently crosswalked to the CPEP inputs of two different CPT codes. We are deleting the crosswalk to the procedure CPT code 21450 and will retain the crosswalk to the anesthesia CPT code 00140 (Anesthesia for procedures on eye).

• We believe that the supply inputs for the retrobulbar injection codes (CPT codes 67500, 67505, and 67515) have been inappropriately crosswalked by the CPEP panel from adjacent surgical procedure codes. After consultation with an ophthalmology specialty society, we have adjusted the supplies so that the list now includes one alcohol swab, one pair of nonsterile gloves, one 5-cc syringe, and one 25-gauge needle.

• In several of the in-office ophthalmology codes, the supply list includes the costs for 50 to 100 sterile towels. The specialty society has confirmed that this is a typographical error and that the quantity should not exceed five for any one visit or procedure. We have made the appropriate adjustments.

• The supply list for CPT code 68761 (Close tear duct opening by plug), currently does not include the costs of a punctal plug. We have received a comment from the specialty society representing optometrists requesting that we add this supply because it is typically used for this procedure. We agree with this comment and are proposing the addition of a punctal plug to the CPEP supplies. We have also deleted the inappropriate inputs from HCPCS code A4263, permanent tear duct plug.

• We have discovered a calculation error that affects the total cost of supplies for some of the codes for which the RUC made recommendations in 1999. We have made the appropriate corrections and are using the corrected values for this rule.

• We have adjusted the clinical staff and supply inputs for HCPCS code G0170, skin biograft, to reflect that it is a 10-day global service with one postprocedure visit.

After consultation with the specialty society, we have also adjusted the supplies for CPT code 53040, drainage of deep periurethral abscess, to correct for anomalies in the quantity of supplies between the in and out of office settings.

(d) Calculation of Practice Expense Pools—Other Issues.

The Act requires payment of some practitioner services (services of certified registered nurse anesthetists, nurse practitioners, clinical nurse specialists, physician assistants, and certified nurse mid-wives) based on a percentage of the physician fee schedule payment amount. Since the payment under the physician fee schedule for a service performed by a midlevel practitioner is required to be based on a percentage of the amount paid to a physician for a service, we are proposing using only physician practice expense data in determining the practice expense RVUs for each practitioner service. Removal of the services performed by midlevel practitioners from the practice expense calculations would assist in simplifying the methodology and would also be consistent with the statutory requirement that we pay for their services based on a percentage of the fee schedule amount.

We have received comments from several surgical specialties urging us to evaluate the Medicare claims data to eliminate potential errors. (For example, claims for non-surgeons performing complex surgeries that are generally performed by surgical specialties only.) These commenters were concerned that incorrect specialty utilization will decrease a specialty's practice expense pool and recommended that these claims should either be reassigned to the appropriate specialty or excluded during refinement. To determine whether potential errors in the claim data have an adverse impact on any specialty or merely represent “noise” that creates no significant effect, we ran the following analyses:

First, we analyzed the utilization for CPT codes 63045 through 63048, the highest volume neurosurgical procedures performed by neurosurgeons. Our utilization data indicates that 91 percent of allowed services for these codes are performed by neurosurgeons and orthopedic surgeons. Of the 9 percent of allowed services when the utilization data indicates another specialty, 3 percent are attributed to general surgeons. An additional 2 percent are attributed to the HCFA specialty code for a clinic or other group practice, when it is likely that a surgeon who is a member of a multispecialty clinic is providing the surgical service. Of the remaining 4 percent of allowed services, the data indicates a specialty of general practice, family practice, or neurology.

For the utilization attributed to general and family practitioners, the data indicate that, in most cases, these physicians are serving as assistants-at-surgery. With respect to neurology (2 percent of the allowed services), we believe it is possible that a physician may practice as both a neurologist and neurosurgeon and designate neurology as the specialty for reporting on Medicare claims. For an insignificant percentage of the allowed services (under 1 percent of the allowed services for all remaining specialties combined), our data indicate a specialty that would not be expected to perform the neurosurgical procedure. In these cases, the incorrect CPT code might have been transcribed on the Medicare claim or the incorrect specialty code may have been reported. There was a similar pattern for services associated with other surgical specialties.

We then tested the impact of reassigning to the dominant specialty this small proportion of allowed services associated with specialties not expected to perform them. We selected three of the specialties that commented on the possibility of erroneous utilization data and identified the complete range of specialized codes associated with each specialty. We reassigned to each dominant specialty the utilization currently assigned to other specialties not expected to perform the services. In addition, to test the “worst-case” scenario, we then crosswalked all frequencies for their complete range of codes to the selected individual specialty.

When we recoded CPT codes 61000 through 64999 to neurosurgery only, the impact on neurosurgery was a 0.55-percent increase. When we recoded the specialty for only those specialties that would not be expected to provide CPT codes 61000 through 64999 (specialties other than neurosurgery, orthopedic surgery, group practice or physician assistant) to neurosurgery, the resulting impact on neurosurgery was a 0.69-percent increase. In reviewing the utilization data for this code range, we found services that are predominantly performed by radiologists and anesthesiologists (such as CPT code 62311). When we recoded only those services predominantly performed by neurosurgeons, the impact was even less.

When we recoded the specialty for all utilization in the range of CPT codes 65091 through 68899 to ophthalmology only, the impact on ophthalmology was 0.31 percent. When we recoded the specialty for only those specialties that would not be expected to provide CPT codes 65091 through 68899 to ophthalmology, the resulting impact on ophthalmology was a 0.32-percent increase.

When we recoded the specialty for all utilization in the range of CPT codes 69000 through 69979 to otolaryngology, the impact on otolaryngology was a −0.36 percent. When we recoded the specialty for only those specialties that would be expected to provide CPT codes 69000 through 69979 to otolaryngology, the resulting impact on otolaryngology was −0.35 percent.

We believe that these simulations exaggerate the potential impact of possible errors in the utilization data because, as discussed in the above analysis of CPT codes 63045 to 63048, our simulations likely reassigned the specialty in situations in which the specialty was correctly coded. In any case, in no scenario did the impacts even approach a 1-percent increase or decrease.

We also believe these simulations demonstrate that the small percentage of potential errors in our very large database have no adverse effect on specialty-specific practice expense RVUs. Therefore, we are not proposing any further action at this time.

The Lewin Group has recently begun the third phase of the project. This phase will concentrate specifically on evaluating the indirect cost allocation methodology. They will evaluate the validity of our current methodology that allocates indirect costs using direct costs and work RVUs and consider alternatives to allocating indirect costs by the current method. The Lewin Group will perform a variety of tasks during this phase of the project to evaluate the advantages and shortcomings of our current indirect cost allocation methodology, as well as of any alternative methodologies. The preliminary tasks for Phase III include—

• Analyzing the current indirect cost allocation methodology to identify its advantages and shortcomings;

• Considering alternate ways in which our methodology might weight direct costs and work RVUs in the allocation of indirect costs and predicting the effects of these alternatives;

• Evaluating the impact and value of changing the methodology to use time rather than work measurements to allocate indirect costs;

• Interviewing experts in the field on potential alternatives to the current indirect cost allocation methodology; and

• Reviewing other relevant efforts to allocate indirect costs associated with physician and non-physician practice expenses.

The Lewin Group's draft final report will present the findings from all three phases of The Lewin Group's analysis of our practice expense methodology. As mentioned above, we are planning to extend The Lewin Group's contract for another year to obtain additional assistance on issues related to practice expense refinement.

(e) Site of Service. Clarifying the Definition of Facility/Nonfacility.

For purposes of practice expense calculations, we make a distinction between services performed in a non-facility and a facility setting. This distinction takes into account the higher expenses of the practitioner in the non-facility setting when the practitioner typically bears the cost of the resources (for example, clinical staff, supplies, and equipment) associated with the services. In the facility setting, because these costs are not incurred by the physician, Medicare payment to the facility includes the cost of the resources for the services furnished. The purpose of the distinction in the site-of-service is to ensure that Medicare does not duplicate payment, to the physician and the facility, for any of the practice expenses incurred in performing a service for a Medicare patient.

For purposes of applying the site-of-service differential, we are defining hospitals, skilled nursing facilities, and ambulatory surgical centers as facilities because they will receive a facility payment for their provision of services. We have been advised that community mental health centers (CMHCs) should also be defined as a facility setting since CMHCs also receive a separate facility payment for their services. Therefore, we are proposing to revise § 414.22(b)(5)(i) (Practice expense RVUs) to add CMHCs to the settings listed in which we would apply the facility practice expense RVUs.

In addition, while we have indicated in previously published rules that the non-facility practice expense RVUs are applicable to outpatient therapy services (physical therapy, occupational therapy, and speech language pathology) furnished by comprehensive outpatient rehabilitation facilities or outpatient rehabilitation providers, there is confusion about this issue. Only the facility can bill for therapy services furnished to hospital and SNF patients. Because this facility payment must include amounts reflecting practice expenses, the higher nonfacility RVUs are used to pay for therapy services even in the facility setting. Therefore, we would amend § 414.22(b)(5)(i) to specifically provide that the nonfacility practice expense RVUs are applicable to outpatient therapy services regardless of the actual setting.

The Act requires that payments vary among fee schedule areas according to the extent that relative costs vary as measured by the GPCIs. Generally, the fee schedule areas that existed under the prior reasonable charge system were retained under the fee schedule from calendar years 1992 to 1996. We implemented a comprehensive revision in fee schedule payment areas (localities) in 1997, reducing the number of localities from 210 to 89. A detailed discussion of fee schedule areas can be found in the July 2, 1996 proposed rule (61 FR 34615) and the November 1996 final rule (61 FR 59494). We are required by section 1848(e)(1)(A) of the Act to develop separate indices to measure relative cost differences among fee schedule areas compared to the national average for each of the three fee schedule components. While requiring that the practice expense and malpractice indices reflect the full relative cost differences, the Act requires that the work index reflect only one-quarter of the relative cost differences compared to the national average.

Section 1848(e)(1)(C) requires us to review and, if necessary, adjust the GPCIs at least every 3 years. This section of the Act also requires us to phase in the adjustment over 2 years and implement only one-half of any adjustment in the first year if more than 1 year has elapsed since the last GPCI revision.

The GPCIs were first implemented in 1992. The first review and revision was implemented in 1995, and the second review was implemented in 1998. This constitutes the third GPCI review and revision and will be implemented in 2001.

The GPCIs were developed by a joint effort of researchers at the Urban Institute and the Center for Health Economics Research under contract to HCFA. Indices were developed that measured the relative cost differences among areas compared to the national average in a “market basket” of goods. In this case, the market basket consists of the resources involved with operating a private medical practice. The resource inputs are physician work or net income; employee wages; office rents; medical equipment, supplies; malpractice insurance; and other miscellaneous expenses. Employee wages, rents, medical equipment, supplies, and other miscellaneous expenses are combined to comprise the practice expense component of the GPCI. The weights of these components in the original GPCIs (from 1992 through 1994), the first (1995 through 1997) and second (1998 through 2000) GPCI revisions, and the new weights for the third proposed GPCI revision (2001 through 2003) are as follows:

The resource inputs and their weights were obtained from the AMA's Socioeconomic Characteristics of Medical Practice Survey. The weights for the 1992 through 1994 GPCIs were from the AMA's 1987 survey, the latest available when the original GPCIs were being developed. The weights for the 1995 through 1997 and 1998 through 2000 GPCIs were from the 1989 survey. The 1989 weights are those used in the revised Medicare Economic Index (MEI) discussed in the November 25, 1992 final rule (Medicare Program; Revision of the Medicare Economic Index) (57 FR 55899). The weights in the proposed 2001 through 2003 GPCIs are from the 1997 AMA survey and were used in the MEI revision discussed in November 2, 1998 final rule (Medicare Program; Revisions to Payment Policies and Adjustments to the Relative Value Units Under the Physician Fee Schedule for Calendar Year 1999) (63 FR 58846).

The MEI is a measure of annual increases in the cost of operating a private medical practice and is used in the annual update of the physician fee schedule CFs. Because the GPCIs and the MEI use the same resource inputs to measure the costs of a private medical practice (the GPCIs measure relative costs among areas while the MEI measures the national annual rate of increase in costs), we believe the same weights should be used.

Once the components and their weights were determined, data sources had to be found that were widely and consistently available in all physician fee schedule areas to measure costs. After examining many sources, the following proxies were selected as the best available sources for measuring each component of the original 1992 through 1994 GPCIs:

• Physician work—The median hourly earnings, based on a 20 percent sample of 1980 census data, of workers in six professional specialty occupation categories (engineers, surveyors, and architects; natural scientists and mathematicians; teachers, counselors, and librarians; social scientists, social workers, and lawyers; registered nurses and pharmacists; writers, artists, and editors) with 5 or more years of college. Adjustments were made to produce a standard occupational mix in each area. The actual reported earnings of physicians were not used to adjust geographical differences in fees because these fees are, in large part, the determinants of the earnings. We believe that the earnings of physicians will vary among areas to the same degree that the earnings of other professionals vary.

• Employee wages—Median hourly wages of clerical workers, registered nurses, licensed practical nurses, and health technicians were also based on a 20-percent sample of 1980 census data.

• Office rents—Residential apartment rental data produced annually by the Department of Housing and Urban Development (HUD) were used because there were insufficient data on commercial rents across all physician fee schedule areas.

• Miscellaneous expenses—The Urban Institute and the Center for Health Economics Research assumed that this component is represented by a national market and that costs do not vary appreciably among areas. This component's index is 1.000 for all areas to indicate no variation from the national average.

• Malpractice—Premiums in 1985 and 1986 for a mature “claims made” policy (a policy that covers malpractice claims made during the covered period) providing $100,000 to $300,000 of coverage were used. Adjustments were made to incorporate the costs of $1 million to $3 million coverage and mandatory patient compensation fund requirements. Premium data were collected for physicians in three risk classes—low-risk (general practitioners who do not perform surgery), moderate risk (general surgeons), and high-risk (orthopedic surgeons).

The areas selected for measurement purposes were the Metropolitan Statistical Areas (MSAs). Non-MSA areas within a State were aggregated into one residual area. Using MSAs for measurement satisfied the criteria of (1) homogeneity in resource input prices within the area, and (2) a size large enough so that market areas are self-contained to minimize border crossing; that is, physicians would not move their offices a few miles to secure higher payments and patients would tend to receive services within their area.

The Act requires, however, that the GPCIs reflect cost differences among fee schedule areas. Thus, it was necessary to map Medicare localities to the MSA and non-MSA aggregation of GPCI data. Where localities crossed MSA boundaries, MSA indices were converted to Medicare locality indices by population weights.

Detailed discussions of the methodology and data sources of the 1992 through 1994 GPCIs can be obtained by requesting the following studies from the National Technical Information Service by calling 1-800-553-NTIS or, for residents of Springfield, Virginia, (703) 487-4650.

• The Urban Institute report “The Geographic Medicare Index: Alternative Approaches,” NTIS PB89-216592.

• The supplement to “The Geographic Medicare Index: Alternative Approaches,” NTIS PB91-113506. This was published in the September 4, 1990 notice for the model fee schedule (55 FR 36238).

• The Urban Institute report, “Refining the Malpractice Geographic Practice Cost Index,” February 1991, NTIS PB91-155218. The related diskette is NTIS PB91-507491. This is the final version of the 1992 through 1994 GPCIs as published in the November 1991 final rule (56 FR 59785).

The main criticism of the original GPCIs was that they were outdated because they were based on old data; for example, 1980 census data and 1985 and 1986 malpractice premiums. This was, however, the most recent data available when the GPCIs were established. The revised 1995 through 1997 GPCIs were based on the most current data available when they were developed in 1993 and 1994.

We made some minor changes from the original GPCI methodology in calculating some of the revised 1995 through 1997 indices. One methodological change was made that applied across all indices. As mentioned earlier, under the original GPCIs, where Medicare localities crossed MSA boundaries, MSA indices were converted to locality indices by population weights. Medicare expenditure weights were not used because the expenditures under the reasonable charge system contained large differences unrelated to relative cost differences among areas. In calculating the revised GPCIs, where localities crossed MSA boundaries, locality indices were calculated by weights based on the proportion of localities' RVUs provided in each MSA to reflect relative cost differences among areas. Full fee schedule RVUS were used rather than actual 1993 payments because 1993 fee schedule payments still reflected some reasonable charge payment levels. The advantages of RVU weighting are (1) the GPCIs more closely reflect physician practice costs in the area where the services are provided rather than where the population lives, and (2) budget neutrality is preserved when we combine multiple localities into larger areas, such as statewide localities.

a. Work Geographic Practice Cost Indices. Data from the 20-percent sample of census data of median hourly earnings for the same six categories of professional specialty occupations as used in the 1992 through 1994 work GPCIs were used in calculating the 1995 through 1997 work GPCIs. The 1992 through 1994 work GPCIs were calculated using 1980 census data of earnings for professionals with 5 or more years of college. That sample was no longer available with the 1990 census. The 1990 census educational classifications were by highest degree earned and not by years of schooling as in the 1980 census. Thus, it was not possible to obtain earnings data that exactly compared to the 1980 data.

For 1990, data were available for all-education and advanced-degree samples, but not for 5 or more years of college. We elected to use the all-education sample because its larger sample sizes made it more stable and accurate in the less populous areas. Although it could be argued that physicians' earnings might more closely approximate the earnings of professionals with advanced degrees, the differences between the all-education and advanced-degree indices were negligible in all but a few of the smallest localities. We believed that the small sample sizes of advanced-degree occupations in these small localities may produce inaccurate results.

The 1992 through 1994 work GPCIs used metropolitan-wide median wages for each county within an MSA. That is, all counties within an MSA were assigned the MSA-wide median wage even if there were wage variations within the MSA. We believed that this was appropriate for all but Consolidated Metropolitan Statistical Areas (CMSAs), the largest of the MSAs, such as New York. In these CMSAs, we replaced metropolitan-wide earnings with county-specific earnings. We believed this change was appropriate because costs were, in fact, higher in central city areas (for example, Manhattan and San Francisco) than in the rest of the CMSA. County earnings better account for cost variation within these large metropolitan areas.

b. Practice Expense Geographic Practice Cost Indices. (1) Employee Wage Indices. Data from the 20-percent sample of census data of median hourly earnings for the same categories of medical and clerical occupations used in the 1992 through 1994 practice expense GPCIs were used in the 1995 through 1997 practice expense GPCIs. The 1995 through 1997 practice expense GPCIs used 1990 rather than 1980 census data. As with the work GPCIs, county level data were used for CMSAs to better reflect the cost variations within these large metropolitan areas.

(2) Rent Indices. As with the original rent indices, the HUD fair market rental (FMR) data for residential rents were again used as the proxy for physician office rents. The 1995 through 1997 practice expense GPCIs reflect 1994 HUD FMRs. Like the work GPCI and the employee wage index of the practice expense GPCIs, county level data were used in CMSAs to recognize the variations within the CMSA.

The major criticism of the rent indices was that residential rather than commercial rent data were used. As mentioned earlier, for constructing the GPCIs, we needed data that were widely and consistently available across all physician fee schedule areas. As with the original GPCIs, we again searched for private sources of commercial rent data that were widely and consistently available.

The private sources we found were not adequate. None of the sources collected data for nonmetropolitan areas, nor did any collect data for all metropolitan areas. The sources did not reflect the average commercial space in the area, but rather the particular type of space most relevant to the needs of a particular source's clients. In addition, the sample sizes were small. A comparison of the average rental for any particular city showed significant variation depending upon the source. Also, the private commercial rent data tended to be for very high priced real estate of the type likely to be used by large institutions such as banks, insurance companies, or financial firms and not for the type of office space used by physicians.

Among the sources of commercial rent data available, the most promising were data from the Building Owners and Managers Association, the General Services Administration, and the U.S. Postal Service. These data were analyzed in depth. We did not use data from the Building Owners and Managers Association and the General Services Administration because of poor geographic coverage, especially outside of large metropolitan areas. That is, data were not widely and consistently available for all physician fee schedule areas. The U.S. Postal Service data had much better geographic coverage, but sample sizes in many areas were unacceptably small and could have led to erroneous results.

No acceptable national commercial rent data are readily available for physician office rents. Thus, some proxy must be used for this portion of the index. In addition, commercial rent data are not available for all areas from published statistical sources.

We believe that the HUD FMR data remain the best available data for constructing the office rental index. They are available for all areas, are updated on an annual basis, and are consistent among areas and from year to year. Moreover, physicians are frequently located in areas and office space that are residential rather than commercial (for example, in apartment complexes and small strip commercial centers adjacent to residential areas).

(3) Medical Equipment, Supplies, and Miscellaneous Expenses. As mentioned earlier, the GPCI assumes that this component has a national market and that input prices do not vary among geographic areas. We were unable to find any data sources that demonstrated price differences by geographic area. Anecdotal and interview data from suppliers and manufacturers were inconclusive. While some price differences may exist, they are more likely to be based on volume discounts rather than on geographic areas. Generally, it appears that manufacturers' prices do not vary among areas except for shipping costs. Since manufacturers and suppliers are located all over the country, shipping costs on the mainland do not vary significantly.

We did consider an add-on for shipping costs to Alaska, Hawaii, and Puerto Rico to recognize the added shipping distance. We decided against the add-on because there were no data to indicate how much the costs of shipping medical equipment and supplies to these areas increased their total costs. We were able to ascertain that commercial shippers like United Parcel Service and Federal Express generally charge about 10 percent more to ship to Puerto Rico and about 20 percent more to ship to Alaska and Hawaii from the mainland.

Medical equipment and supplies represent about 7 percent of physician practice costs. Even assuming that shipping costs represent 5 percent of total equipment and supply costs, which we believe to be a high estimate, recognizing a 20 percent increase in shipping costs would only increase payment levels by 0.07 percent or 0.0007 (.20 × .05 × .07 = .0007). The medical equipment, supplies, and miscellaneous expense index for all areas continued to be 1.000 in the revised 1995 through 1997 GPCIs.

c. Malpractice Geographic Practice Cost Indices. Again, malpractice premium data for a $1 million to $3 million mature “claims made” policy were collected, with mandatory patient compensation funds considered. However, more recent and more comprehensive malpractice insurance data were used in calculating the 1995 through 1997 malpractice GPCIs. The 1995 through 1997 malpractice GPCIs were based on 1990 through 1992 premium data. Malpractice premiums are very volatile and may change significantly from year to year. We decided to use the most recent 3-year average available rather than just the most recent single year to smooth out this volatility and present a more accurate indication of malpractice premium trends over time.

We collected data on more specialties and from more insurers. We collected data on 20 specialties, rather than on only three as in the 1992 through 1994 malpractice GPCIs. The 1992 through 1994 malpractice GPCI data were largely drawn from a single nationwide insurer (St. Paul Fire and Marine) and were supplemented by several State-specific carriers in States in which St. Paul did not offer coverage. Subsequent analyses suggest that these data were not representative of insurers operating in many States. For the revised malpractice GPCI, data were collected from insurers that, on average, represented 82 percent of the market in each State, with the lowest State market share being 60 percent. We believe that the more recent and much more comprehensive data greatly improved the accuracy of the malpractice GPCIs for 1995 through 1997.

Detailed discussions of the methodology and data sources of the 1995 through 1997 GPCIs can be obtained by requesting the following studies from NTIS by calling 1-800-553-NTIS, or (703) 487-4650 in Springfield, Virginia:

• “Updating the Geographic Practice Cost Index: Revised Cost Shares.” Debra A. Dayhoff, John E. Schneider, and Gregory C. Pope. NTIS PB94-161072.

• “Updating the Geographic Practice Cost Index: The Physician Work GPCI.” Gregory C. Pope and Deborah A. Dayhoff. NTIS PB94-161080.

• “Updating the Geographic Practice Cost Index: The Practice Expense GPCI.” Gregory C. Pope, Deborah A. Dayhoff, Angella R. Merrill, and Killard W. Adamache. NTIS PB94-161098.

• “Updating the Geographic Practice Cost Index: The Malpractice GPCI.” Stephen Zuckerman and Stephen Norton. NTIS PB94-161106.

The same data sources and methodology used for the 1995 through 1997 GPCIs were used for the revised 1998 through 2000 GPCIs with a few very minor modifications. No acceptable additional data sources were found. The cost shares were the same as in the 1995 through 1997 GPCIs because no changes were made in the MEI weights.

Indices for fee schedule areas are based on the indices for the individual counties within the fee schedule area. Fee schedule RVUs are again used to weight the county indices (to reflect volumes of services within counties) when mapping to fee schedule areas and in constructing the national average indices. However, we used more recent data, 1994 rather than 1992 RVUs, in the county, locality, and national mapping in the proposed GPCIs. The payment effect of this is negligible in most cases and generally results in changes at the third decimal point if at all.

a. Work Geographic Practice Cost Indices. The work GPCIs are based on the decennial census. The 1992 through 1994 work GPCIs were based on 1980 census data because 1990 census data were not yet available. The work GPCIs were revised in 1995 with new data from the 1990 census. New census data will not be available again until after the 2000 census. We searched for other data that would enable us to update the work GPCIs between the decennial censuses but no acceptable data sources were found. The most promising sources of data were the hospital wage data that we collected to calculate the prospective payment system (PPS) hospital wage index and the payroll per worker data collected by the U.S. Bureau of Labor Statistics from State unemployment insurance agencies (“the ES-202 data”).

The PPS hospital wage data were examined when we constructed the original GPCIs. They were rejected in favor of census data because of their lack of an occupation mix adjustment and their unrepresentative occupational composition (hospital employees rather than professionals or physician office employees). ES-202 data consist of total payroll divided by counts of wage and salary workers. Their major disadvantages were that they did not measure hourly earnings, only payroll per employee, and no occupational detail is available. Also, they did not adjust for part-time or full-time and hours worked, and the numbers of workers are small for certain States, leading to unstable estimates of payroll per worker. We compared the changes by State from 1989 to 1993 in the PPS wage data and the ES-202 data to see if there was any correlation between the two series. The correlation between the two was only moderate: 0.55. The changes indicated by both series were generally small, for example, a few percentage points. The difference between the two series by State was in many cases as large as, or greater than, the change indicated by either series. The average difference between the two series (2.1 percent) is as large as the change indicated by either series. In addition, changes for particular States were substantially different between the two series. For example, Indiana relative wages rose by 1.9 percent according to the PPS data, but fell 5.7 percent according to the ES-202 data.

Since we were unable to find an acceptable data source for updating the work GPCIs, we examined the consequences of not updating the work GPCIs between the decennial censuses. We compared the changes between the 1992 through 1994 work GPCIs, based on the 1980 census and the 1995 through 1997 GPCIs, based on the 1990 census. On average, the full variation State work GPCIs changed by about 5 percent. This translates to about a 1.2 percent change in the quarter work GPCI required by law. Since work makes up about one-half of the GPCI cost shares, this translates into an average payment change per State of about 0.6 percent from updating the work GPCI based on the 10-year change in relative wages indicated by the census data. Even the maximum change in the full variation State work GPCIs from the 1992 through 1994 to the 1995 through 1997 GPCIs of 14 percent translates into only about a 1.8 percent change in payments. The largest full work GPCI changes for individual payment areas were from 16 to 20 percent, or about a 4 to 5 percent change in the quarter work GPCI, or about a 2.4 percent change in payments. However, 80 percent of payment areas experienced payment changes of less than 1 percent, and 50 percent of payment localities experienced payment changes of less than 0.5 percent as a result of changes in the census data from 1980 to 1990.

We, therefore, made no changes in the 1998 through 2000 work GPCIs from the 1995 through 1997 work GPCIs, other than the generally negligible changes resulting from using 1994, rather than 1992, RVUs for this GPCI update because we were unable to find acceptable data for use between the decennial censuses. We believe that making no changes is preferable to making inaccurate changes based on unacceptable data. We believe that this is a reasonable position given the generally small magnitude of the changes in payments resulting from the changes in the work GPCIs from the 1980 to the 1990 census data.

b. Practice Expense Geographic Practice Cost Indices. (1) Employee Wage Indices. As with the work GPCIs, the employee wage portion of the practice expense GPCIs is based on decennial census data. For the same reasons discussed above pertaining to the work GPCIs, we made no changes in the employee wage indices during the 1998 through 2000 GPCI update. The average change from the 1992 through 1994 to the 1995 through 1997 employee wage indices across States was about 6 percent. Since the employee wage index had a weight of about 16 percent in the GPCI cost shares, this translated into a 1 percent average change in payments. The maximum payment change in any payment area resulting from changes from the 1992 through 1994 to the 1995 through 1997 employee wage indices was about 3.2 percent. Payment changes in over two-thirds of the payment areas were less than 1 percent.

(2) Rent Indices. The office rental indices were again based on HUD residential rent data. The rental indices were based on 1996 HUD data as opposed to the 1994 HUD data in the 1995 through 1997 GPCIs. HUD made two small methodological changes in developing the data. First, HUD used the 40th percentile of area rents rather than the 45th percentile. This did not materially affect the GPCIs, which measure relative rents among areas. Second, HUD established a rental floor for rural counties at the statewide rural average. This had the effect of raising the office rental indices slightly in rural areas.

We made one methodological change in the rent indices. HUD publishes FMRs only for metropolitan areas as a whole. For the 1995 through 1997 GPCIs, HUD used a special tabulation of the 1990 census data to allocate rents by county within CMSAs. In some metropolitan areas, this had the effect of reducing the central city index below the suburban index, probably because of lower unmeasured housing quality in central cities than in suburbs. This may not have been the best indicator of relative physician rents, since the GPCIs are intended to measure rental costs for offices of similar quality in different areas. The metropolitan-wide rent is most appropriate for measuring the cost of space of an average quality across the metropolitan area, which is why HUD publishes only metropolitan-wide FMRs. Also, the census county adjustments can be updated only once every 10 years. For this reason, we believed that the county-specific adjustment should not be made for all large metropolitan areas, but should be retained only for the New York City Primary MSA. Available evidence suggested that rents vary substantially among the boroughs of New York City and that, given the current locality configuration, the county-specific rental adjustment appropriately reflected these patterns in the New York City area, especially the higher rents in Manhattan.

(3) Medical Equipment, Supplies, and Miscellaneous Expenses. As with the 1992 through 1994 and 1995 through 1997 GPCIs, this component was given a national value of 1.000, indicating no measurable difference among areas in costs.

c. Malpractice Geographic Practice Cost Indices. Again, malpractice premium data were collected for a mature “claims made” policy with $1 million to $3 million limits of coverage, with adjustments made for mandatory patient compensation funds. As with the 1995 through 1997 GPCIs, data were collected for the 20 largest Medicare-billing physician specialties. The premium data represent at least 50 percent of the market in each State. Again, we used an average of the 3 most recent premium years to smooth out the considerable year-to-year fluctuations that can occur in malpractice premiums. The revised 1998 through 2000 malpractice indices were based on 1992 through 1994 premium data, the latest years available when the Health Economics Research (HER) GPCI study was being conducted in 1995 through 1996. Another change from the 1995 through 1997 indices is that we weighted the specialty shares of the 20 specialties by fee schedule RVUs rather than allowed charges.

Detailed discussions of the methodology and data sources of the 1998 through 2000 GPCIs may be obtained by requesting the following study from NTIS by calling 1-800-533-NTIS, or, for residents of Springfield, Virginia, (703) 487-4650: “Second Update of the Geographic Practice Cost Index.” Gregory C. Pope and Killard W. Adamache.

We propose using the same data sources and methodology used for the 1998 through 2000 GPCIs for the 2001 through 2003 GPCIs (hereafter referred to as proposed GPCIs). No acceptable additional data sources were found. The only differences between the 1998 through 2000 GPCIs and the proposed GPCIs are in the cost shares and RVU weighting. As shown in the cost share table in the discussion of the development of the GPCIs, the cost shares have been changed to reflect the revisions in the MEI. This does not affect the work or malpractice GPCIs since they are stand-alone indices. The change has a small effect on the practice expense GPCIs because it changes slightly the weights among the employee wage, rents and miscellaneous components of the practice expense index. We used more recent RVU data—1998 rather than 1994—in the county, locality, and national mapping in the proposed GPCIs. The payment effect of this is generally negligible.

a. Work Geographic Practice Cost Indices. For the same reasons discussed in the section on the 1998 through 2000 work GPCIs, no significant changes are being proposed in the 2001 through 2003 work GPCIs because we were unable to find acceptable data for use between the decennial censuses. There are general negligible changes resulting from the use of 1998 rather than 1994 RVUs for weighting.

b. Practice Expense Geographic Practice Cost Indices. (1) Employee Wage Indices. As with the work GPCIs, the employee wage indices are based on decennial census data. For the same reasons discussed above pertaining to the work GPCIs, we are proposing no changes in the employee wage indices during this GPCI update.

(2) Rent Indices. The office rental indices are again based on HUD residential rent data. No changes have been made in the methodology. The proposed rental indices are based on 2000 rather than 1994 HUD data.

The proposed rental indices are compared to the current rental indices in Addendum D. A reduction in an area's rent index does not necessarily mean that rents have gone down in that area since the last GPCI update. Since the GPCIs measure area costs compared to the national average, a decrease in an area's rent index means that an area's rental costs have decreased when compared to the change in national average rental costs. The indices are arranged in descending order of change. The rental index has a cost share of about 12 percent of the GPCI. This means that the actual effect on payments will be about 12 percent of the change in the rental indices. While the new rental indices show significant changes in a few areas, primarily in the San Francisco Bay area, 80 of the 89 areas change by less than 10 percent, which translates into about a 1 percent change in payments.

(3) Medical Equipment, Supplies, and Miscellaneous Expenses. As with all previous GPCIs, this component would be given a national value of 1.000, indicating no measurable differences among areas in costs.

c. Malpractice Geographic Practice Cost Indices. We propose using the same methodology described in the 1998 through 2000 malpractice GPCI section in the proposed malpractice GPCIs for 2001 through 2003. The only difference is that we used more recent data. The proposed malpractice indices are based on 1996 through 1998 data compared to the 1992 through 1994 data used in the previous GPCI update.

Addendum E shows the changes from the 1998 through 2000 indices to the proposed malpractice GPCIs. A change in an area's malpractice GPCI does not mean that absolute malpractice premiums have changed by that amount. It, rather, reflects the area's new position compared to the national average. As with past GPCI revisions, the changes in the proposed malpractice GPCIs are relatively large in some cases, reflecting the significant changes in malpractice premiums that occur from year to year. As Addendum E shows, two-thirds of the payment areas experience changes of less than 12 percent. It should be noted, however, that the weight of the malpractice GPCI is only about 3 percent of the total GPCI. Therefore, a 12 percent change in the malpractice GPCI translates into only a 0.4 percent change in payments. Even the largest 42 percent change in the malpractice GPCI translates into only a 1.3 percent change in payments. The mean change in the malpractice GPCIs is 11 percent, or about a 0.4 percent change in payments.

The proposed 2002 fully-effective revised GPCIs and the transitional 2001 revised GPCIs can be found at Addendum F and Addendum G, respectively. Since the proposed revised GPCIs could result in total payments either greater or less than payments that would have been made if the GPCIs were not revised, it was necessary to adjust the GPCIs for budget neutrality as required by law. Therefore, we adjusted the 2001 through 2002 GPCIs as follows: work by 0.99699; practice expense by 0.99235; and malpractice by 1.00215.

In the July 1999 proposed rule (64 FR 39610) and the November 1999 final rule (64 FR 59383) for the CY 2000 physician fee schedule, we discussed the methodology used to calculate resource based malpractice RVUs and proposed interim RVUs effective January 1, 2000. (See “Legislative History” section for dates and Federal Register citations for these rules.) The methodology can be briefly summarized as follows:

• Actual malpractice premium data were collected for the top 20 Medicare physician specialties.

• All Medicare specialties were mapped to insurer rating classes (ISO codes).

• A national average premium was calculated for every specialty.

• Specialty risk factors showing the relative malpractice costs among specialties were created by dividing each specialty national average premium by the lowest average premium.

• Specialty-weighted malpractice RVUs were calculated for each procedure by summing, for all specialties providing the procedure, the product of each specialty's risk factor times the proportion of total service count for that procedure provided by the specialty.

• This number was multiplied by the procedure's work RVUs to account for differences in risk-of-service among procedures.

• The new malpractice RVUs were adjusted by the appropriate factor to attain budget neutrality.

The malpractice RVUs were based on 1993 through 1995 premium data, the most recent premium data readily available. In last year's proposed and final rules we stated that we planned to collect more recent data, but did not expect that newer data would change the values significantly since malpractice premiums have been remarkably stable in recent years.

We have now obtained, and are currently examining, malpractice premium data for 1996 through 1998. The malpractice RVUs in the fall final rule will reflect the newer data. While we have not yet completed the proposed malpractice RVU calculations, the table below compares the 1993 through 1995 average premiums (that were used to calculate the 2000 malpractice RVUs) with the 1996 through 1998 average premiums (that will be used to calculate the 2001 malpractice RVUs). As the table below shows, there was very little change in the national average premiums from 1993 through 1995 to 1996 through 1998. We, therefore, anticipate minimal changes in malpractice RVUs from use of the more recent data.

In addition, in response to comments received on last year's rules, we are proposing to accept a comment regarding crosswalking specialties. We are proposing to crosswalk surgical oncology to general surgery rather than to all physicians. The malpractice values to be included in the final rule reflecting the updated data will remain interim.

In the November 1999 final rule (64 FR 59423), we established interim work RVUs for CPT codes 99291 and 99292 (critical care services) of 3.6 and 1.8, respectively, which were decreased from the previous RVUs for these services. These work RVUs were established because of the change in the CPT definition of critical care services in CPT 2000. We also discussed in detail what changes in the definition most concerned us. We received many comments on the interim work RVUs for critical care.

This year we proposed new coding language to the AMA CPT Editorial Panel (the Panel) to resolve physician concerns. The Panel, with input from various specialty societies, accepted the language that we proposed with some modifications. The AMA has given us copyright permission to publish the introduction for CPT codes 99291 and 99292 as it will appear in CPT 2001. For CPT 2001, the introduction for critical care services will be as follows (new language in italics):

The remainder of the introduction as published in CPT 2000, as well as the descriptors for the two CPT codes (99290 and 99291), remains unchanged.

Adoption of this revised introduction for the critical care CPT codes 99291 and 99292 is consistent with our view of the appropriate intensity of these services and addresses the concerns we had raised in the November 1999 final rule. Therefore, based on implementation of this revised introduction for critical care services for CY 2001, we are proposing to value the physician work at 4.0 RVUs for CPT code 99291 and 2.0 RVUs for CPT code 99292.

In addition, consistent with our discussion in the proposal for electrical bioimpedance (EB) (see section II.H), we are proposing to not allow separate Medicare payment for EB when provided in conjunction with critical care services (CPT codes 99291 and 99292).

The Panel considered changes to the definition of care plan oversight for 2001. After analyzing the definition changes, we are concerned that these codes (CPT codes 99375 and 99378) will no longer be consistent with our coverage criteria.

In anticipation of the likely CPT revisions, we would establish two new HCPCS codes for care plan oversight that are consistent with our coverage criteria. For the 2001 physician fee schedule, we would establish a new HCPCS code Gxxx1, that will use the CPT 2000 definition associated with CPT code 99375 and a new HCPCS code Gxxx2, that will use the CPT 2000 definition associated with CPT code 99378. The current policy guidance that applied to CPT codes 99375 and 99378, including our past responses to questions on care plan oversight, will continue to apply to these G codes. The current payments for CPT codes 99375 and 99378 will be maintained in Gxxx1 and Gxxx2.

In addition, we would establish two new HCPCS codes (Gxxx3 and Gxxx4) to describe the services involved in physician certification (and recertification) and development of a plan of care for a patient for whom the physician has prescribed Medicare-covered home health services. The proposed text of the new codes will read as follows:

This code would be used when the patient has not received Medicare-covered home health services for at least 60 days.

This code would be used after a patient has received services for at least 60 days (or one certification period) when the physician signs the certification after the initial certification period.

The use of these HCPCS codes (Gxxx3 and Gxxx4) would be restricted to physicians who are permitted to certify that home health services are required by a patient pursuant to section 1814(a)(2)(C) and section 1835(a)(2)(A) of the Act. The Gxxx3 code would be billed only once every 60 days, except in the rare situation when the patient starts a new episode before 60 days elapses and requires a new plan of care to start a new episode. Consistent with section 1835(a)(2) of the Act, a physician who has a significant ownership interest in, or a significant financial or contractual relationship with a home health agency (HHA), generally cannot bill this code for patients served by that HHA.

For services within the episode (generally beyond the first week or two of care plan implementation) that are consistent with the definition of care plan oversight (HCPCS code Gxxx1), the care plan oversight code (CPT code 99375) would be used.

Because we believe that the physician work associated with HCPCS code Gxxx3 equates to that of a level 3 established patient office visit (CPT code 99213), we are proposing a value of .67 for the work RVUs. For Gxxx4, because we believe the work equates to a level 2 established patient office visit (CPT code 99212), we are proposing a value of .45 for the work RVUs. For practice expense RVUs, we are proposing to crosswalk both Gxxx3 and Gxxx4 to the practice expense inputs currently used for care plan oversight (CPT code 99375).

In 1998, the AMA added new CPT codes 99234 to 99236, Observation or inpatient hospital care services (including the admission and discharge services) for a patient on the same date. We accepted the RUC recommendations for work RVUs for these new codes. The work RVUs for each code are the sum of the applicable admission work for CPT codes 99218 to 99220 (or CPT codes 99221 to 99223) plus the discharge work (CPT codes 99217 or 99238). For example, CPT code 99234 has 2.56 work RVUs, which is the sum of the work RVUs for CPT code 99221 (1.28) plus the work RVUs for CPT code 99217 (1.28). However, it has come to our attention that allowing payment for these CPT codes conflicts with two policies currently in the Medicare Carrier Manual (MCM).

Section 15505.1(c) of the MCM states that we will pay for only the initial hospital care service code when a patient is admitted as an inpatient and discharged on the same day. Physicians are not paid for both an inpatient hospital admission and hospital discharge management on the same day. In addition, section 15504.b of the MCM instructs that CPT codes 99218 to 99220 (Initial observation care) should be used if the patient is discharged on the same day as the admission for observation because each of these codes represents a full day of care and, thus, paying for a code representing both admission and discharge on the same day would be duplicative. CPT code 99217 (Observation care discharge) may be billed only on the second or subsequent days in observation.

These two payment policies result in different payments for patients whose inpatient stay is less than 24 hours based solely on whether they were in the hospital at midnight. For example, a physician who admits a patient to observation or to inpatient care at 8 a.m. and then discharges the patient at 8 p.m. the same day, would be allowed payment for only the admission service. On the other hand, a physician who admits a patient to observation or to inpatient care at 8 p.m. and then discharges the patient at 8 a.m. the next day, would be allowed payment for both the admission and discharge services.

In response to these concerns, and to clarify our payment policy, we are proposing the following:

Inpatient stay of 24 hours or more —We would pay for both inpatient hospital admission services (CPT codes 99221 to 99223) and hospital discharge services (CPT codes 99238 to 99239) when a patient is a hospital inpatient for a period of 24 hours or more. The medical record must document that the patient was an inpatient for at least 24 hours for both of these services to be paid.

Inpatient or observation stay of less than 8 hours —If a patient is admitted as a hospital inpatient or an observation patient for less than 8 hours, we would pay for only the admission service (CPT codes 99221 to 99223 or 99218 to 99220) on that day. The discharge service is not considered to be a separately billable service.

Inpatient or observation stay of 8 or more hours, but less than 24 hours —If a patient is admitted as a hospital inpatient or an observation patient for a period of 8 or more hours, but less than 24 hours, we would pay for both the admission and discharge services under CPT codes 99234 to 99236 with the following proposed physician work RVUs and documentation requirements:

Physician Work RVUs —To properly value both the admission and discharge work of these services, we are proposing to continue valuing the admission portion of the physician work as equivalent to CPT codes 99218 to 99220 (or CPT codes 99221 to 99223), but to reduce the discharge work RVUs from 1.28 to 0.67. This would make the discharge portion of the work equal to the work for CPT code 99213 (Office or other outpatient visits) instead of CPT code 99217 (or CPT code 99238). Thus, the proposed work RVUs would be as follows: CPT code 99234—1.95 RVUs; CPT code 99235—2.81 RVUs; CPT code 99236—3.66 RVUs. We would not pay CPT codes 99217, 99238, and 99239 for hospital inpatient or observation admissions between 8 and 24 hours in length.

Our reasoning for these proposed RVUs is that we believe that the physician work typically required for discharging an inpatient or observation admission patient after a period of at least 8 hours, but less than 24 hours, is less than that required for an admission of 24 hours or more. The typical work (for example, history, physical examination, and medical decision making) and the typical face to face time required to discharge such a patient is comparable to the requirements for CPT code 99213. Moreover, the typical time for CPT code 99238 is up to 30 minutes and the physician work is 1.28 RVUs, so a clear work anomaly would be created if we made the work value of discharging a patient with a stay of less than 24 hours identical to the work of discharging a patient with a length of stay of 24 hours or more.

Our proposal would avoid creating such a rank order anomaly and would place admission and discharge valuation in proper order. For example, for observation stays of less than 8 hours, we would pay only the admission portion and would not pay separately for the discharge because the extra work is minimal. For observation stays of more than 8 hours, but less than 24 hours, we would recognize the discharge component since there is significant extra work involved, but not as much as a discharge for a 24 hour or longer admission for which we would pay the full value of CPT code 99238. Our proposal would allow payment for CPT codes 99234 through 99236 only for stays of equal to or greater than 8 hours, but less than 24 hours.

In addition to the documentation guidelines for history, physical examination, and medical decision making described in CPT 2000 for CPT codes 99234 to 99236, we would require the following to be documented in the medical record:

• A stay involving 8 hours, but less than 24 hours.

• That the billing physician was present and personally performed the services.

• Admission and discharge notes written by the billing physician.

We believe this policy would harmonize current policy on hospital admissions and discharges and also accommodate the observation codes as they are described in CPT 2000. The policy would not be tied to the “midnight” time frame of the hospital inpatient census.

If these proposals are adopted in the final rule, the work RVUs for CPT codes 99234 to 99236 would be considered interim for 2001.

Ocular photodynamic therapy is a treatment recently approved by the Food and Drug Administration for age-related macular degeneration, the most common cause of blindness in the elderly. For CPT 2000, ocular photodynamic therapy was added to CPT code 67220, which was formerly limited to photocoagulation by laser.

We believe that ocular photodynamic therapy is significantly different from laser photocoagulation and, therefore, we are proposing to establish new HCPCS codes that specifically identify these procedures. A discussion of each of these codes follows:

This code would be used in place of CPT code 67220. We would maintain the work and malpractice RVUs and the CPEP inputs presently used for CPT code 67220 for payment of this new “G” code.

We are proposing a value of 0.55 work RVUs for Gxxx6. This value is half the physician work value for CPT code 96570 (Photodynamic therapy by endoscopic application of light to ablate abnormal tissue via activation of photosensitive drug(s); first 30 minutes), and it is identical to the physician work value for CPT code 96571 (Photodynamic therapy by endoscopic application of light to ablate abnormal tissue via activation of photosensitive drug(s); each additional 15 minutes). We note that the total time of laser light application for ocular photodynamic therapy is 83 seconds, which is considerably shorter than the time of laser light application for CPT codes 96570 and 96571.

We are also proposing that the global period for Gxxx6 be “XXX.” Because of the global designation, significant, separately identifiable evaluation and management (E/M) services may be billed on the same day as Gxxx6 with the use of the -25 modifier. Patients will, typically, have fluorescein angiography as well as an E/M service before ocular photodynamic therapy to determine whether they will benefit from the therapy and to discuss the treatment. Any E/M services performed after the treatment may be billed separately.

For Gxxx6 we are proposing the following practice expense inputs for non-facility settings:

• Clinical Staff Time: Registered nurse/ophthalmology technician—40 minutes.

• Supplies: Ophthaine, mydriacil, myolfrin, gonisol, post myd spectacles, verteporfin and also infusion supplies including sterile and non-sterile gloves, butterfly needle, syringe, band aid, alcohol swab, staff gown, iv infusion set, and infusion pump cassette.

• Equipment: Laser, infusion pump, and exam lane.

For the malpractice component of Gxxx6, we are proposing 0.52 RVUs (the value assigned to CPT code 67220, Destruction of localized lesion of choroid). Although we are establishing procedure codes for ocular photodynamic therapy, coverage of the procedure is at the discretion of the local carrier.

In instances where both eyes are treated the same day, we are proposing the use of the following HCPCS add-on code:

For this add-on code we are proposing a “ZZZ” global period, with .28 work RVUs (half of that proposed for Gxxx6) and .52 malpractice RVUs (identical to that proposed for Gxxx6). The practice expense inputs for services in the non-facility setting would be as follows:

• Clinical Staff Time: Registered nurse/ophthalmology technician—5 minutes.

• Supplies: Ophthaine, mydriacil, myolfrin, gonisol.

In addition, we have identified several other specific ophthalmological treatments that are not distinctly identified in CPT 2000. We are proposing to establish specific HCPCS codes for these procedures.

We are not proposing RVUs for HCPCS codes Gxxx8 through Gxx10. These codes are being established for tracking purposes only. These procedures are considered experimental in nature at this time and, therefore, are not covered under Medicare.

Electrical bioimpedance (EB), a noninvasive method of measuring cardiac input, is a covered procedure under Medicare, if medically necessary. Performance of this procedure is reported by the Level 2 HCPCS code M0302, and the procedure is currently carrier priced. We are proposing the following RVUs for this procedure:

We are proposing the following direct inputs for determining practice expense RVUs. (We note, however, that a final determination of the practice expense RVUS will depend on how we value physician work.) The practice expense RVU in Addendum B reflects the value for the technical portion of the service. If the service is given physician work, a separate PC will be established with an additional practice expense RVU.

• Clinical staff: Registered nurse—15 minutes.

• Supplies: Four disposable sensors, patient gown, exam table paper, and pillowcase.

• Equipment: Cardiac output monitor and exam table.

We are proposing 0.02 RVUs for this procedure. This value is equivalent to the TC of an EKG, which is a similar procedure.

The uses for which this procedure are covered (for example, differentiating cardiogenic from pulmonary causes of acute dyspnea, the need for intravenous inotropic therapy, fluid management, and the uses indicated in section 50-54 of the Coverage Issues Manual, HCFA Pub. 6) require a clinical evaluation of the patient on the same day that EB is performed. The procedure reports measurements that can not be interpreted without other clinical information.

With respect to proposed RVUs for physician work, we have insufficient information to propose a work value. We are collecting information and invite comments on this subject as well as on the proposed inputs for practice expense and malpractice. In your comments, please be sure to compare your proposed value for the physician work component for this service to other similar services with established physician work values. Please also include the reason why you believe the physician work is similar. At this time, we have received comments proposing no physician work values, proposing physician work values similar to that for the interpretation of an EKG (CPT code 93010—0.17 work RVUs), proposing work values similar to total body plethysmography (CPT code 93720—0.17 work RVUs), and similar to interpretation of cardiovascular stress test (CPT code 93018—0.30 work RVUs).

We also are proposing that the payment for this procedure be included in reporting critical care. Therefore, separate payment would not be made for this procedure when provided in conjunction with critical care services (CPT codes 99291 and 99292).

Currently, there is a 90-day global period in the physician fee schedule for all CPT codes involving the insertion, removal, and replacement of pacemakers or cardioverter defibrillators. During the global surgical period, no separate payment may be made for any E/M service furnished by the surgeon, unless the visit is: (1) Unrelated to the diagnosis for which the surgical procedure was performed; (2) for treating the underlying condition; or (3) an added course of treatment that is not part of normal recovery from surgery.

In these situations, the surgeon must use CPT modifier -24 that attests that the E/M service provided, although performed during the postoperative period, was for a reason unrelated to the original procedure. Services submitted with a -24 modifier must be sufficiently documented to establish that the visit was unrelated to the surgery. An ICD-9-CM code that clearly indicates that the reason for the encounter was unrelated to the surgery is acceptable documentation.

Many patients receiving pacemakers or cardioverter defibrillators have clinically serious cardiac diseases (related to the reason for the procedure) that require significant postoperative care. In these cases, it is difficult to separate care during the postoperative period for the related cardiac problem(s) from the postoperative care for the pacemaker or cardioverter defibrillator procedure. As medical practice has changed, cardiologists predominantly perform pacemaker or cardioverter defibrillator procedures. Thus, the physician performing the pacemaker or cardioverter defibrillator procedure now is typically the same physician who is expected to furnish care for the patient's related cardiac disease. Therefore, a single physician is providing postoperative care for both the pacemaker or cardioverter defibrillator insertion and the related medical problem(s), but can be paid only for the insertion because of the global period policy.

We believe it is common for patients undergoing pacemaker and cardioverter defibrillator procedures to require significant care for related cardiac disease during the postoperative period. This care overlaps substantially with the care furnished for the pacemaker or cardioverter defibrillator procedure and may be coded with the same ICD-9-CM diagnosis code; therefore, using the -24 modifier is inadequate to allow appropriate payment for the physician performing both postoperative care and care for the patient's other cardiac conditions.

We are proposing to change the global period for CPT codes 33206, 33207, 33208, 33212, 33213, 33214, 33216, 33217, 33218, 33220, 33233, 33234, 33235, 33240, 33241, 33244, 33249, 33282, and 33284 from 90 days to 0 days. This would permit separate payment for any care furnished during the postoperative period by the physician who performed the pacemaker or cardioverter defibrillator procedure.

We are soliciting comments on whether it is appropriate to reduce the global period for these CPT codes. We are also proposing to ask the RUC to revise the RVUs for these CPT codes. If RUC recommendations are not received in time for our consideration for the CY 2001 physician fee schedule final rule, we propose to implement interim work RVUs, as listed below.

In calculating the proposed interim RVUs, we have subtracted the work RVUs of all postoperative visits after the day of surgery from the total work RVUs. We used our database to calculate the number of postoperative visits. Where our database did not contain the number of postoperative visits, we crosswalked a number from the most clinically similar procedure. We have included an example to illustrate the calculation.

Example: For CPT code 33206, the 2000 work relative value is 6.67 units. The proposed work value is 3.11 (6.67 minus 3.56). The 3.56 units represents the work based on the pattern of E/M services in the global period.

We would also adjust practice expense inputs for supplies, staff time, and equipment to account for the change in the global period. Because these would be 0-day global services only priced in the facility setting, there would be no direct CPEP inputs associated with them. The adjusted practice expense RVUs are reflected in Addendum B.

We welcome comments on our proposed calculation of interim RVUs and request that commenters recommending RVUs include the methodology employed so that we can appropriately evaluate the recommended RVUs. As an alternative to applying a 0-day global period as discussed above, we are interested in other suggestions that might address the issue of assuring appropriate payment for these services (for example, adjusting the global period to 10 days for these services). We invite public comment on such alternatives.

Section 410.68(b), Antigens: Scope and conditions, provides for beneficiaries to receive a supply of antigen for no more than 12 weeks at one time. A specialty society has indicated that this limitation is not reflective of current industry standards and guidelines (for example, duration of potency for allergy extracts has changed since the policy was implemented.) Therefore, we are proposing to change this limitation from 12 weeks to 12 months and would revise the regulations to reflect this change. We are requesting comments on this proposal.

In the November 1999 (64 FR 59419) final rule, we assigned RVUs to CPT code 20979, low intensity ultrasound stimulation to aid bone healing. Commenters expressed concern about the RVUs assigned to this service. Because of the concerns raised by commenters, and because CPT code 20979 is a noncovered service under Medicare, we are proposing to remove the RVUs that were assigned to this code at this time. We may reconsider this at a future date.

In the April 11, 2000 correction notice (65 FR 19332) to the November 1999 final rule, we inadvertently published practice expense RVUs based on the work RVUs associated with a 90-day global period for CPT codes 33975 and 33976 (implantation of ventricular assist devices). However, in the same notice, the global periods and associated work RVUs for CPT codes 33975 and 33976 were revised to reflect an “XXX” (the global concept does not apply). In calculating the practice expense RVUs, we reflected changes made in CPEP data that result from changes in the global period. However, the practice expense RVUs are also a function, in part, of the physician work RVUs. In calculating the revised practice expense RVUs, we did not use the work RVUs that reflected the global period change. Effective January 1, 2001, we would revise the practice expense RVUs associated with these CPT codes to reflect the revision in the global periods and work RVUs.

Under current policy, medical supervision by an anesthesiologist occurs if the anesthesiologist is involved in furnishing more than four concurrent procedures or is performing other services while directing fewer than four concurrent procedures. Payment is based on three base units plus one unit for induction if the physician is present at induction.

Under current policy, medical direction by an anesthesiologist occurs if the anesthesiologist is involved in two to four concurrent anesthesia procedures or a single anesthesia procedure with a qualified anesthetist. For each anesthesia procedure, the anesthesiologist must—

• Perform a pre-anesthesia examination and evaluation;

• Prescribe the anesthesia plan;

• Personally participate in the most demanding procedures of the anesthesia plan, including emergence and induction;

• Ensure that any procedures in the anesthesia plan that he or she does not perform are performed by a qualified anesthetist;

• Monitor the course of anesthesia administration at frequent intervals;

• Remain physically present and available for immediate diagnosis and treatment of emergencies; and

• Provide indicated post anesthesia care.

We currently do not have a national policy that instructs the carriers how to pay for a service when the anesthesiologist does not fulfill all the medical direction requirements. One option carriers may use is to instruct the anesthesiologist to report this service as a reduced or unusual service and determine appropriate payment. We are considering clarifying this policy and making other revisions to the medical supervision payment policy. We are considering the following:

1. To specify that the physician furnishing medical supervision must perform, at a minimum, the preoperative evaluation, participate in induction, remain available for consultation, and provide a minimum level of monitoring.

2. To establish payment for medical supervision at 40 percent of the payment amount for the service performed by the physician alone.

3. To apply the proposed medical supervision payment amounts to incompletely medically-directed cases.

4. To limit the number of concurrent cases the physician can supervise to five concurrent cases.

Payment for medical supervision is payment for the physician service. In addition, the certified registered nurse anesthetist (CRNA) service furnished under medical supervision is paid at 50 percent of the amount that would have been paid if the service had been performed by the physician alone.

We invite comments from the public, but in particular, the physicians and practitioners most affected by this policy. We are not proposing a change at this time, but will consider the comments we receive should we develop a future proposal.

In the November 1999 final rule (64 FR 59413), we indicated that we would adopt our proposal to bundle payment for certain diagnostic codes, including pulse oximetry CPT codes 94760 and 94761, into the payment for other services. We believe that continuing to pay separately for these codes duplicates amounts included in both facility payments and practice expense RVUs. However, we did not address how we would treat situations when these services are performed without any other billable service and, thus, are not reflected in facility payments or other practice expense RVUs. We will continue to pay separately for these services (CPT codes 94760 and 94761) when they are medically necessary and there are no other services payable under the physician fee schedule billed on the same date by the same supplier.

In the November 1998 final rule (63 FR 58868), we stated that we were maintaining our current requirement that therapy assistants of therapists in private practice (formerly known as therapists in independent practice (PTIP)) must be personally supervised by the therapist and be employed directly by the therapist; employed by the partnership or group to which the therapist belongs; or employed by the same practice. Personal supervision requires that the therapist be in the same room during the performance of the service. Levels of supervision are defined at § 410.32 (Diagnostic X-ray tests, diagnostic laboratory, and other diagnostic tests: Conditions.)

The November 1998 final rule did not change pre-existing regulations at § 410.60(c)(2) (Supervision of physical therapy services) for therapy assistants in a private practice setting. In that final rule, we codified the statutory requirements for coverage of outpatient occupational therapy services by establishing § 410.59 (Outpatient occupational therapy services: Conditions). Section 410.59 parallels the requirements in § 410.60 for outpatient physical therapy, as revised in the November 1998 final rule. We also made conforming changes in § 410.61 (Plan of treatment requirements for outpatient rehabilitation services) to include occupational therapy.

The personal supervision requirements for therapy assistants and aides in a private practice setting are long-standing. The outpatient physical therapy benefit, enacted in 1972, applied to PTIP (that is, individual therapists in independent practice in their own offices). Services performed by employees of a PTIP were covered if furnished under the direct personal supervision of the PTIP. This requirement was necessary to assure beneficiary health and safety and quality of care.

In 1981, in response to the conference committee report (H.R. 96-1479) accompanying the Omnibus Reconciliation Act of 1980 (Public Law 96-499), we revised our Medicare Carriers Manual instructions (see section 2215F, HCFA Pub. 6). These revised instructions stated that the services of employees of a PTIP who are not qualified physical therapists must be furnished under the direct personal supervision of a supervising therapist who must be the employer or on the employer's staff. Therefore, a licensed physical therapist had to directly and personally supervise the services of assistants and aides. Thus, even before the November 1998 final rule, the regulations and manuals clearly stated that the PTIP must directly and personally supervise all services for which he or she bills.

As noted above, pre-existing supervision requirements for therapy assistants in a private practice setting were not affected by the November 1998 final rule. However, we received comments from the therapy industry and other interested parties who erroneously believed that we had either misinterpreted the supervision requirement or had established a new requirement for therapy assistants in the private practice setting.

These comments and the confusion possibly resulted from the one revision in supervision requirements made in the final rule. This revision related not to therapy assistants, but to qualified therapists in a private practice setting. As referenced in the November 1998 final rule (63 FR 58868), the Congress was concerned about the requirement for therapists in independent practice to directly supervise all services performed by their employees, even if those employees were fully-licensed therapists. The therapist in independent practice had to be on the premises whenever services were furnished to Medicare beneficiaries, including services furnished by a licensed therapist. Therefore, a therapist in independent practice could not have more than one office open at the same time because he or she could not be on both premises at once. Congressional statements in both the House and Senate committee reports associated with our fiscal year 1997 appropriations process addressed this issue. The House committee report urged us to modify the regulations so that certified therapists need not be on the premises to supervise other licensed therapists. We were also urged by the Senate to review this concern and recommend changes in our regulations or instructions. To address this concern expressed in both the 1997 House and Senate Appropriations Committee reports, we revised the regulations at § 410.59(c)(2) and § 410.60(c)(2).

Accordingly, effective January 1, 1999, as specified in the November 1998 final rule, the revised regulations permit legally authorized (see § 410.59(c)(1)(i) and § 410.60(c)(1)(i)) therapists who own the practice to be off the premises when other legally authorized therapists are present to furnish supervision for therapy assistants. These regulations also restated which practitioners are qualified as therapists under section 1861(p) of the Act. In accordance with the November 1998 final rule, the term “independent” was removed from the description of a therapist in independent practice. In its place, the term “private” was added. The benefit is now described in terms of an individual physical therapist or occupational therapist in private practice.

This change did not affect the required degree of supervision for physical therapist assistants. Assistants still must be personally supervised by the therapist in private practice and employed directly by the therapist, by the partnership, or group to which the therapist belongs.

Section 221 of the BBRA placed a 2-year moratorium on Medicare Part B outpatient therapy caps (the $1500 cap on outpatient physical therapy services including speech language-pathology services and the $1500 cap on outpatient occupational therapy services in all nonhospital settings). The two $1500 caps were implemented in 1999 as required by the BBA.

The BBRA also requires us to submit to the Congress a report by January 1, 2001 that includes recommendations on—(1) the establishment of a mechanism for assuring appropriate utilization of outpatient therapy services; (2) the establishment of an alternative payment policy for these services based on classifications of individuals by diagnostic category, functional status, prior use of services (in both inpatient and outpatient settings), and other criteria, in place of uniform dollar limitations; and (3) how to do this in a budget-neutral manner.

We are gathering information on alternatives or options that we can use to achieve these objectives. We have received the following informal recommendations for legislation:

• Institute a cap per diagnosis rather than per year.

• Establish payment based on patient groupings by primary diagnosis and average number of treatments, with options for variants.

• Base payment on an episode of occurrence of illness or injury, with a cap amount adjusted to address geographic differences in the cost of furnishing services.

• Develop a sustainable growth rate (SGR) for outpatient therapy services to control growth in the volume of services.

The outpatient therapy cap was also a topic of discussion at the PPAC meeting in December 1999. As a result of these discussions, the PPAC recommended continuation of the current moratorium with focused medical review, indicating that such a review could lead to the desired budget-neutral outcome.

We would like comments from the public on additional alternatives that we could consider in developing a payment policy for outpatient therapy services. We will consider this information as we prepare our report to the Congress on outpatient therapy services.

Section 1848(c)(2)(B)(i) of the Act requires that we review all RVUs for services in the physician fee schedule no less often than every 5 years. The first 5-year review was undertaken as part of the final rule published December 1994 (59 FR 63140), with the resulting changes effective for services furnished beginning January 1, 1997. In the final rule published November 1999 (64 FR 59427), we included a discussion of the first 5-year refinement and outlined our plans for the second 5-year refinement of work RVUs. We also solicited comments on potentially misvalued work RVUs as well as data sources and methodologies to assist us in identifying misvalued services. We received comments from approximately 30 specialty groups, organizations, and individuals. While some of the comments were on the proposed process, comments also included requests for evaluating over 900 codes.

As we had discussed in the November 1999 final rule, in addition to performing internal review and analysis, we will be sharing these comments with the RUC, which currently makes recommendations to us on the assignment of RVUs to new and revised CPT codes. The RUC's perspective will be helpful because of its experience in recommending RVUs for the codes that have been added to the CPT, or revised by the CPT panel, since we implemented the physician fee schedule in 1992. We emphasize, however, as we reiterated for the first 5-year review, we have the responsibility for analyzing the comments concerning the 5-year review and deciding whether to revise RVUs. We are not delegating this responsibility to the RUC or any other organization.

Current initiatives underway (see discussion that follows on physician time) will assist us in our identification of misvalued codes. However, we will not be able to identify those codes that we believe have misvalued work RVUs before late in the year 2000 or early in the year 2001. We propose to perform the 5-year review in two phases. The first phase will take place in CY 2000 with consideration of public comments. The second phase will occur in CY 2001 when we use the contracted research to identify misvalued codes. We will work with the RUC and the medical community to minimize work duplication. For example, we will ask the RUC to defer action in CY 2000 on those codes that were identified by public comments and that our research later indicates might be misvalued. Furthermore, to focus on each phase of the review and prevent duplicative work, we propose to concentrate on intraspecialty issues and anomalies in CY 2000 and consider cross specialty misvaluations and issues in CY 2001. This is because we believe that validation of time across a wide range of services will allow direct comparison of pre-, intra-, and postservice work RVUs across specialties with the potential to identify a large number of misvalued codes. Again, we will work closely with the medical community to analyze and interpret the data as well as to organize the review in an efficient manner.

We currently have initiatives underway to validate the physician time data and identify potentially misvalued codes to be considered during the 5-year review. A discussion of these activities follows.

Under a contract with HCFA, Health Economics Research (HER) is reviewing secondary data sources to validate time estimates for physicians' services. Physician time estimates are a factor used in the calculation of the practice expense RVUs and one of the primary determinates of physician work. These secondary data sources are as follows:

• The National Ambulatory Medical Care Survey (NAMCS).

• D.J. Sullivan Associates Hospital Data.

• MGMA Practice Cost Survey Data.

The NAMCS is a survey conducted by the Center for Disease Control that collects self-reported information on over 20,000 office visits annually including physician face-to-face time (called the duration of the visit). Various comparative analyses, both at the physician specialty level and for all physicians, can be made between projected E/M codes in the NAMCS data and with the actual E/M codes reported in the Medicare Part B National Claims History. (E/M codes are not captured in the NAMCS data. However, a method is used to map the time of the physician visit to an appropriate E/M code. This represents the “projected” E/M code.) The analysis was performed on the 1997 NAMCS.

The D.J. Sullivan database groups approximately 495,000 inpatient and outpatient records into 177 small clinically similar classes. The database captures information from the hospital record such as the procedure, time the patient enters the operating room, time of incision, time of wound closure, and time the patient exits the operating room. Data are presented for all hospitals and for all hospitals by categories: community hospitals, teaching hospitals, and university-based hospitals.

HER is analyzing a sample of the D.J. Sullivan database to determine whether it can be used for validating skin-to-skin time for selected surgical procedures. The selected procedures are high volume procedures or procedures on the RUC multispecialty list.

The MGMA Physician Profiling Database contains information at the physician practice level on the number of services by CPT code, physician specialty, and clinical work week. The database contains information on almost 4,000 physicians, primarily from Florida, Minnesota, New York, and Washington. Analysis will focus on comparing expected clinical times based on current time estimates attributable to CPT codes to total practice hours worked.

This document does not impose information collection and recordkeeping requirements. Consequently, it need not be reviewed by the Office of Management and Budget under the authority of the Paperwork Reduction Act of 1995.

Because of the large number of items of correspondence we normally receive on Federal Register documents published for comment, we are not able to acknowledge or respond to them individually. We will consider all comments we receive by the date and time specified in the DATES section of this preamble, and we will respond to the major comments in the final rule.

We have examined the impacts of this proposed rule as required by Executive Order of 1993 (EO) 12866, the Unfunded Mandates Reform Act of 1995 (EO) 12875 (UMRA) (Public Law 104-4), the Regulatory Flexibility Act of 1980 (RFA) (Public Law 96-354) and the Federalism Executive Order of 1999 (EO) 13132.

EO 12866 directs agencies to assess costs and benefits of available regulatory alternatives and, when regulation is necessary, to select regulatory approaches that maximize net benefits (including potential economic, environmental, public health and safety effects, distributive impacts, and equity). A regulatory impact analysis (RIA) must be prepared for major rules with economically significant effects ($100 million or more annually). While the changes in the Medicare physician fee schedule are for the most part, budget neutral, they do involve redistribution of Medicare spending among procedures and physician specialties and geographic areas. However, the redistributive effect of this rule on any particular specialty or geographic area is, in our estimate, unlikely to exceed $100 million. The effect of the practice expense changes are estimated to increase payments to one specialty by about $90 million and decrease payments to another specialty by approximately $45 million. All other physician specialties will be affected by less than these amounts. The GPCI changes are expected to increase payments by less than $10 million in one locality and decrease payments by about $20 million in another locality. The effect on all other payment localities are likely to be less than these amounts. Since we estimate that these changes are unlikely to redistribute more than $100 million in Medicare allowed charges, we are not considering this proposed rule to be a major rule. However, we will reconsider this decision for the final rule if our estimates based on new data exceed $100 million.

The UMRA also requires (in section 202) that agencies prepare an assessment of anticipated costs and benefits before developing any rule that may result in expenditure in any one year by State, local, or tribal governments, in the aggregate, or by the private sector, of $100 million or more. We have determined that this proposed rule will have no consequential effect on State, local, or tribal governments. We believe the private sector cost of this rule falls below the above stated threshold as well.

The RFA requires that we analyze regulatory options for small businesses and other small entities. We prepare a Regulatory Flexibility Analysis unless we certify that a rule would not have a significant economic impact on a substantial number of small entities. The analysis must include a justification of why action is being taken, the kinds and number of small entities the rule affects, and an explanation of any meaningful options that achieve the objectives and lessen significant adverse economic impact on the small entities.

In addition, section 1102(b) of the Act requires us to prepare a regulatory impact analysis if a rule may have a significant impact on the operations of a substantial number of small rural hospitals. This analysis must conform to the provisions of section 603 of the RFA. For purposes of section 1102(b) of the Act, we define a small rural hospital as a hospital that is located outside of a Metropolitan Statistical Area and has fewer than 50 beds.

For purposes of the RFA, all physicians are considered to be small entities. There are about 700,000 physicians and other practitioners who receive Medicare payment under the physician fee schedule. We have prepared the following analysis, which, together with the rest of this preamble, meets all four assessment requirements. It explains the rationale for and purpose of the rule, details the costs and benefits of the rule, analyzes alternatives, and presents the measures we considered to minimize the burden on small entities.

Revisions in resource-based practice expense RVUs for physicians' services are calculated to be budget neutral, that is, the total practice expense RVUs for calendar year 2001 are calculated to be the same as the total practice expense RVUs that we estimate would have occurred without the changes proposed in this regulation. This also means that increases in practice expense RVUs for some services will necessarily be offset by corresponding decreases in values for other services.

Table 1 shows the impact on total allowed charges by specialty of this proposed rule's practice expense changes. There are six changes that we made that have on effect on payment for practice expenses. We show the impact of each individual provision and the combined impact on the practice expense RVUs.