Government Affairs Home > Teaching Physicians > Fee Schedule & Other Payment Issues

MedPAC Letter to HCFA on Revisions to Payment Policies Under the Physician Fee Schedule for Calendar Year 1999

August 21, 1998

Nancy-Ann Min DeParle, Administrator
Health Care Financing Administration
Department of Health and Human Services
Attention: HCFA-1006-P
P. O. Box 26688
Baltimore, MD 21207-0488

Re: File Code HCFA-1006-P

Dear Ms. DeParle:

The Medicare Payment Advisory Commission (MedPAC) welcomes the opportunity to comment on the Secretary of Health and Human Services' proposed rule entitled Medicare Program; Revisions to Payment Policies Under the Physician Fee Schedule for Calendar Year 1999, 63 Fed. Reg. 30818 (June 5, 1998). MedPAC commends you and your staff for your efforts to develop the practice expense relative value units (RVUs) in the proposed rule. HCFA staff undertook a large effort to collect data, develop the values, and meet with physicians and other stakeholders and to explain the rationale for the practice expense methodology and its technical details. The Commission appreciates the clear presentation of HCFA's methodology in the proposed rule.

In reviewing the proposed rule, the Commission has focused on the top-down approach HCFA used to develop the proposed RVUs. HCFA's development of a new approach is understandable given the complex issues that arose after last year's proposed rule was released. The Commission agrees that use of this approach is necessary in the near term because of limitations of the direct cost data from the Clinical Practice Expert Panel (CPEP) process. By basing practice expense RVUs on physicians' aggregate practice costs, the top-down approach provides assurance that the RVUs reflect all practice costs, including any that were omitted during the CPEP process.

Implementation of the top-down approach requires the resolution of important data issues during the coming year. Many of these issues involve the survey data available on aggregate physician practice costs. Others pertain to refinement of the CPEP data used to allocate aggregate practice costs to specific services. The Commission has prepared comments on these issues.

Aggregate Practice Cost Data

In the absence of routinely-collected cost data (analogous to facility cost reports), HCFA must rely on surveys for data on physicians' aggregate practice costs. HCFA has used the American Medical Association's (AMA's) Socioeconomic Monitoring System (SMS) survey, a high-quality physician survey that represents the best available source of data for this purpose.

As discussed in the proposed rule, the AMA has identified several problems with the SMS data that should be considered when those data are used to develop practice expense RVUs. First, sample sizes for some specialties are too small to permit separate calculation of their practice costs. Even among specialties with larger sample sizes, variation in the practice expense data means that average practice costs are subject to sampling error. Second, response rates for the practice expense questions in the SMS survey tend to be low relative to other questions. These low response rates may be a source of nonresponse bias. Third, since the SMS survey is a physician-level survey, physicians in groups are asked for their share of practice costs rather than the costs of the entire practice. As noted by the AMA, collection of practice-level data, instead of physician-level data, may reduce measurement error in the estimates of aggregate practice costs. The Commission has considered the different types of potential error--sampling, nonresponse, and measurement error--and their implications for development of practice expense RVUs.

Sampling Error. The data used to calculate the practice expense RVUs in the proposed rule clearly illustrate the relationship between the number of physicians surveyed in each specialty and the sampling error in the estimates of aggregate practice costs (see table). Sampling error is generally greatest for those specialties with the fewest physicians surveyed.

Reducing this sampling error will require the collection of additional data and use of those data to revise and update the RVUs, as necessary. As discussed in the proposed rule, additional data are needed for the smaller specialties not adequately represented in the SMS survey. The rule also notes that additional data on larger specialties may also be used if HCFA receives compelling evidence that the SMS data for those specialties are incorrect. The five-year review of all RVUs, required by the Omnibus Budget Reconciliation Act of 1989, will present another opportunity for collection of aggregate practice cost data for review of the practice expense RVUs. Through such data collection efforts, the precision of the aggregate practice cost estimates for each specialty should improve.

As explained in the proposed rule, HCFA's options for collecting additional practice cost data are: 1) work with the AMA to collect the data as part of the SMS survey, and 2) develop an alternative source of data. The Commission believes these options are not mutually exclusive. HCFA should consider complementary data collection strategies that reduce sampling error and address the nonresponse error and measurement error issues discussed below.

Nonresponse Error. The response rate for the SMS survey's practice expense questions is about 40 percent¹. This response rate is low and could be important, since nonresponse error is a potential source of bias in survey data.

As noted in the proposed rule, major efforts have been made to reduce nonresponse error in the SMS survey data. The AMA makes a pledge of confidentiality to survey participants. To help physicians prepare to answer the practice cost questions, they receive a summary of the practice cost questions prior to the survey. During the survey interview, physicians may designate a proxy, such as a practice manager or accountant, to answer the practice cost questions. As described in SMS survey publications, intensive efforts are made to accommodate physicians' busy schedules and to follow up with those who initially refuse to participate. After the survey data are collected, they are weighted for nonresponse. The weighting procedure accounts for known characteristics of the nonrespondents, including specialty, years since receipt of medical degree, AMA membership status, and board certification status.

Despite these efforts, some potential remains for nonresponse bias in the data. Publications summarizing the SMS data note that the "busyness" of some physicians may explain the low levels of response to the practice cost questions. AMA representatives also report that physicians in larger practices appear to be less likely to respond to the SMS survey's practice cost questions than physicians in smaller practices. Given the use of specialty-specific practice cost data for the proposed rule, any patterns of nonresponse that are related to physician specialty could bias the practice expense RVUs in favor of some services.

For more to be learned about the importance of nonresponse error, data will be needed for an analysis of nonresponse error that compares respondents with non-respondents. One approach to collecting these data would be to amend current survey procedures to allow for more intensive follow-up with physicians who would otherwise not participate in the SMS survey. For example, physicians could be given the option of completing a survey form to be returned by mail. This option may be attractive for physicians in busy or large practices. Information collected by surveyors who visit physician practices may also be an option.

Measurement error. As noted earlier, the AMA has identified one potential source of measurement error in the SMS data: physicians in groups are asked for their share of practice costs rather than the costs of the entire practice. If physicians cannot estimate their share of practice costs accurately, those costs will be measured with error. This source of measurement error will no longer be an issue if the AMA proceeds with plans it is considering to implement a two-year cycle for the SMS survey. During each cycle, a physician-level survey would be fielded in the first year and a practice-level survey would be fielded in the second year.

Other sources of measurement error should also be considered, however. For example, collection of detailed information on practice costs by means of a survey may be a source of measurement error. While physicians participating in the SMS survey receive a summary of the practice cost questions in advance and have an opportunity to prepare for the interview, no mechanism exists to verify the information provided.

As mentioned in the proposed rule, HCFA is considering whether some confirmation of the aggregate practice cost data, through audit or other means, is necessary. If data are used to determine Medicare's physician payments, the Commission believes some procedure for verification of those data would be appropriate, at least to determine the extent to which measurement error is an issue. The Commission believes that an "audit" of the data is not advisable, however. Implementation of audit procedures, in conjunction with a practice cost survey, would further reduce already-low response rates. Other methods for assessing measurement error in the data should be considered.

Collection of Additional Practice Cost Data. From the above discussion, it is clear that collection of additional practice cost data involves achievement of competing goals. Sample sizes need to increase, particularly for the smaller specialties. At the same time, additional data should be collected that allow analyses of nonresponse error and measurement error.

Collection of the additional data must be accomplished carefully, however. The AMA has worked hard to maintain the SMS survey's response rates, and those response rates should not be jeopardized. If the SMS survey cannot serve as a source of data for analyses of nonresponse error and measurement error, the Commission recommends development of an another data source.

The analyses of nonresponse error and measurement error described earlier require changes from current SMS survey procedures more than changes in the content of the survey. One option for addressing these issues might be implementation of alternative procedures for a subsample of SMS survey participants. The Commission encourages HCFA to explore this option with the AMA.

Refinement of Direct Cost Data

Refinement of the direct cost data from the CPEPs is also important. Because the practice expense RVUs for services provided by more than one specialty are a weighted average of specialty-specific RVUs, errors in one specialty's direct cost data can affect other specialties. Errors in the direct cost data are also important in cases where services are provided by only one specialty. Practice expense RVUs that are too high or too low, because of errors in the direct cost data, may create incentives or disincentives for physicians to provide different services.

MedPAC recommends an important role for the RUC, or an RUC-like organization, in the refinement process, instead of continued efforts by HCFA to convene panels of physicians and other professionals to address CPEP data issues. HCFA made good faith efforts to refine the CPEP data last fall, but greater participation and leadership is needed from the physician community. The RUC has proven effective in providing HCFA with recommended work RVUs for new services. A prominent role in the refinement of the CPEP data should prove equally effective.

Conclusion

The Commission's comments address some of the initial steps necessary to implement resource-based practice expense RVUs. Other issues, such as refinement of the physician time and utilization data used in the top-down approach, will become apparent as HCFA receives comments on the proposed rule from physician specialty societies and others. The Commission is developing plans to do further work on these issues and may develop additional recommendations as it fulfills its mission to advise the Congress on Medicare payment policy.

One issue the Commission is likely to address in the future concerns the volume and intensity adjustment described in the proposed rule. In its March 1998 report to the Congress, the Commission recommended against use of such an adjustment and recommended use of the sustainable growth system to make adjustments to the fee schedule's conversion factor if there are volume responses to the new practice expense RVUs. As explained in the proposed rule, HCFA has determined that a volume and intensity adjustment is still necessary based on an analysis conducted by HCFA actuaries. The Commission intends to review the analysis when it becomes available and will address this issue further at that time.

Sincerely,

The overall response rate for the SMS survey is about 60 percent. Of those physicians who participate, about 60 percent answer the survey's practice expense questions.

Sampling Error in Aggregate Practice Cost Data, by Specialty

Source: Medicare Payment Advisory Commission analysis of aggregate practice expense data in Health Care Financing Administration, 63 Fed. Reg. 30830-30831, (June 5, 1998).