Statement of Glenn M. Hackbarth, Chairman, Medicare Payment Advisory Commission

Testimony Before the Subcommittee on Health
of the House Committee on Ways and Means

Hearing on Medicare's Geographic Cost Adjustments

July 23, 2002

Chairman Johnson, Mr. Stark, Members of the Subcommittee. I am Glenn Hackbarth, chairman of the Medicare Payment Advisory Commission (MedPAC). I am pleased to be here this morning to discuss MedPAC's views on how Medicare's payment systems account for differences in local market prices for the goods and services providers must buy to furnish care.

Medicare's payments for services in the traditional program

In the traditional fee-for-service program, Medicare generally uses prospective payment systems (PPSs) to set market-like prices that are intended to encourage efficient delivery of health care services to its beneficiaries. Two of these systems--the PPS for acute inpatient hospital care and the physician fee schedule--are mature systems that have been in place for over a decade. New systems are being phased in for care furnished by hospital outpatient departments, home health agencies, skilled nursing facilities, rehabilitation hospitals, and starting soon, long-term care hospitals.

To ensure access to care for Medicare beneficiaries without imposing undue costs on taxpayers, these payment systems should set payment rates that approximate the costs that efficient providers would incur in furnishing high quality care. Efficient providers' costs will vary because of local market factors--such as prices for labor--that are beyond their control. Consequently, Medicare's payment rates must vary to account for such factors or risk creating undesirable financial incentives and payment inequities.

Adjusting for local market conditions

Market input prices for labor and supplies vary widely across the nation. These input-price differences have substantial effects on providers' costs but are largely beyond their control. Consequently, Medicare's payment rates in each market should be adjusted to reflect the local price level.

How to make these adjustments accurately is one of the most important problems for payment system design and operation. Because input-price differences can account for one-third or more of the variation in unit production costs among providers, errors in input-price adjustments can result in payment inequities and undesirable financial incentives.

Medicare's prospective payment systems generally address this problem by establishing a national base payment rate and then adjusting the rate for the expected relative costliness of the specific case or service and for the local input-price level where the service is furnished. To carry out this design, policymakers must have one or more measures of geographic variation in input prices--such as the area wage index in the acute inpatient hospital care PPS or the geographic practice cost indexes in the physician fee schedule. Policymakers also must know what portions of providers' unit costs are affected by variations in input prices. This information is used to determine how much of the national base payment rate should be adjusted by the geographic input price factor for each market area. Most Medicare payment systems use a version of the hospital wage-index.

The hospital wage index

Medicare's prospective payment systems for inpatient (and other facility) services include input-price adjustments that raise or lower payment rates to reflect the hourly wages of health care workers in each local market, as measured by the hospital wage index. The Centers for Medicare&Medicaid Services (CMS) constructs the hospital wage index for each market area using compensation data from annual hospital cost reports filed by the hospitals located in the area. By law, CMS must define market areas using the 325 Metropolitan Statistical Areas (MSAs) designated by the Office of Management and Budget and 49 statewide rural areas for counties not included in MSAs. The wage index for 2002 varies from a high of 1.53 in Oakland, California to a low of 0.74 for providers in rural Alabama (Figure 1). To address inequities in labor market definitions, particularly for rural hospitals located near the edges of MSAs, Medicare policy allows acute care hospitals to apply for reclassification from one market area to another for the wage index under certain conditions. In FY 2001, 490 hospitals (about 10 percent of all acute care hospitals) were reclassified (Figure 2).

Wage index issues

MedPAC and others have identified four problems with the hospital wage index. ⁽¹⁾   One, the so-called occupational mix problem, in which differences among areas in the mix of workers employed affect the average wage rate, distorting the measurement of market prices for labor. Second, market areas as defined by MSAs and statewide rural areas can be too large, encompassing more than one distinct health care labor market. Third, the wage data that underlie the adjustment are four years old. Finally, the share of the payment to which the input price adjustment is made may include cost components--for example, billing services--that may be purchased in regional or national markets (and whose prices, therefore, should not vary with local market wages).

The effect of differences in the mix of occupations across labor market areas. The objective of the geographic adjustor is to account for differences beyond the control of the provider--local market prices--and not for differences created by management decisions--the mix of labor. Thus, using aggregate wages and hours may distort the wage index by elevating the average wage per hour in markets (such as urban areas with large teaching hospitals) where providers employ a costly mix of labor and depressing the average wage in markets (such as many rural areas) where hospitals employ a relatively inexpensive labor mix. These inaccuracies in the wage index may have substantial effects on payment accuracy. Addressing the occupational mix problem directly will require occupation-specific data that CMS has not yet begun to collect. In the meantime, MedPAC recommended that the Secretary accelerate the planned phase-out from the hospital wage index of salaries and hours for teaching physicians, residents, and certified registered nurse anesthetists. Although the impact would not be large, this policy would improve the distribution of payments. CMS incorporated this suggestion in its proposed rule for hospital payments during fiscal year 2003. We also believe that CMS should collect occupation-specific data on wages and hours using hospitals' annual Medicare Cost Reports, as is done for the aggregate wage and hour data needed to construct the current wage index.

Labor market size. MSAs and statewide rural areas are frequently too large to capture homogeneous labor markets for health care workers. Research has shown a strong pattern of systematic differences in hospital wage levels within many urban and rural labor market areas. ⁽²⁾ Hospitals in outlying suburban counties generally appear to face lower market wage rates than those located in the central core of the same MSA. Similarly, hospitals located in outlying rural areas appear to face lower wage rates than those located in counties adjacent to MSAs. In addition, MSA and state boundaries often separate nearby hospitals (and give them substantially different wage index adjustments) although they are obviously competing in the same labor market. As I mentioned earlier, the Congress established the geographic reclassification policy to ameliorate this boundary problem. But inequities within large market areas remain.

These problems are difficult to resolve, in part because developing consistent criteria that can be used to define labor market areas is technically very difficult. A further barrier, however, is that any change in market definitions creates financial benefits for some providers and financial disadvantages for others, thereby generating great political resistance to reform.

Timeliness of CMS wage data. By the time the wage index is applied to adjust payments, the underlying wage data are four years old. Although the age of the data has often been cited as an important problem, recent research (Dalton et al. 2000) suggests that relative wage levels across geographic areas do not change much over time. Occupation-specific wage data (when available) will allow a more thorough investigation of this issue.

Proportion of costs affected by locally purchased inputs. We also recommend that the Secretary reevaluate current assumptions about the proportions of providers' costs that reflect resources purchased in local and national markets. This so-called labor share estimate is developed and periodically revised by the Office of the Actuary in CMS. The labor share is based on the weights for certain components (categories of inputs) of the hospital market basket index--a measure of annual inflation in the prices of goods and services hospitals buy to produce health care services, which is used in determining annual updates for Medicare's PPS payment rates. Some have argued that the current labor share overstates the proportion of costs that rural hospitals devote to labor and other locally purchased inputs. The components included in the labor share were originally designated in 1983, and many of these are still largely purchased in local markets. However, other inputs may be purchased wholly or partly in national markets, and including them overstates the labor share to some extent. Applying the wage index adjustment using an overstated labor share would lead to underpayment in low-wage areas and over-payment in high-wage areas. For fiscal year 2003, CMS proposes increasing the labor-related share of hospital costs used to apply the wage index from 71.1 percent to 72.5 percent. But analysis sponsored by the Commission indicates that the labor share is at least modestly lower than that currently used, not higher.

The limitations of the hospital wage index have led some advocates to propose that a floor be put under the index. This would raise payments in market areas with low hospital wage rates (and, if done budget neutrally, lower them in areas with high wage rates), but it would do so in an arbitrary fashion. Moreover, if the objective is to help hospitals with poor financial performance, a wage index floor is a poor way to do so because it would raise payments to both low- and high-margin hospitals. Our analysis shows that there is no correlation between hospitals' Medicare inpatient margins and the wage index; hospitals with low margins are just as likely to be located in areas that have high wage indexes as they are to be in areas that have low wage indexes.

1. For a more detailed discussion of wage index issues please see Medicare Payment Advisory Commission, Report to the Congress: Medicare Payment Policy, March 2001.

2. See Dalton, K., Slifkin, R.T., Howard, H.A. Rural hospital area wages and the PPS wage index: 1990 - 1997, available at http://www.shepscenter.unc.edu/research_programs/Rural_Program/wp.html.