USING APPROPRIATE PRICE INDICES FOR ANALYSES OF HEALTH CARE EXPENDITURES OR INCOME ACROSS MULTIPLE YEARS

This document provides guidelines to help ensure consistency and avoid confusion about the use of price indices with MEPS expenditure or income data.
Table 1 contains a summary of the recommended applications of different indices described in this document while Tables 2 and 3 provide the actual price indices for 1996 (first MEPS survey year) through the most recent year available.¹

Table 1: Crosswalk of Price Indices and MEPS Analyses

^1. See http://www.bea.gov/national/#gdp for more information on the Gross Domestic Product (GDP) or Personal Consumption Expenditures (PCE).

^2. See http://www.bls.gov/bls/proghome.htm for more information on the Consumer Price Index (CPI).

^3. See http://www.cms.gov/NationalHealthExpendData/downloads/dsm-09.pdf for more information on the Personal Health Care Expenditure (PHCE) component of the National Health Expenditure Accounts.

1 Comparing Total or Out-of-Pocket Expenditures for Different Years

1.1 Total Expenditures

Most often, comparisons are made of aggregate expenditures of one type or another. As a simple example, total expenditures for the U.S. civilian non-institutionalized population were $553 billion in 1997 and $811 billion in 2002 (both in nominal terms). Comparisons may also be done on a per capita basis, for certain subpopulations, or for certain types of services or drugs (for example, antidepressants). Regardless, the same considerations and options for adjusting expenditures to make cross-year comparisons apply.

• Guideline 1: Make comparisons in nominal (or unadjusted) dollars. That is, leave expenditure estimates as they are, but document this clearly.
• Guideline 2 (recommended): Use the Gross Domestic Product (GDP) price index¹ to put in constant dollars (also known as real or inflation adjusted dollars). For example, to express the 1997 total expenditure estimate in 2002 real (inflation adjusted) dollars, an analyst should multiply the 1997 estimate of $553 billion by a factor of (92.113/84.554) (obtained from Table 2). When compared to actual expenditures in 2002 of $811 billion, this inflated expenditure estimate ($602 billion) indicates that about $209 billion more was spent on health care in 2002 than 1997, after accounting for inflation.

Explanation: In making these comparisons, what we are really asking is how has the amount of society’s resources devoted to health care changed over time. One option is to simply make expenditure comparisons in nominal terms but this does not account for price inflation—a dollar today is not worth a dollar tomorrow. A general price index should be used to put dollar figures into constant terms, because it provides an overall sense of what a dollar can buy today vs. yesterday, be it health care or something else.

While the Consumer Price Index (CPI) is the better known general index, the GDP price index is preferable for a number of reasons. Most notably, the composite CPI only covers about 60 percent of the economy, omitting rural consumers, government purchases, and investment goods. Given the high proportion of health care expenditures that comes from federal, state, and local governments, it is especially important to use a price index, such as the GDP index, that is broadly reflective of the entire U.S. economy. The GDP price index is also used by the Center for Medicare and Medicaid Services (CMS) in the National Health Expenditure Accounts (NHEA).

1.2 Out-of-Pocket Expenditures

• Guideline: Use the Consumer Price Index (CPI all-item all urban consumers) or alternatively the Personal Consumption Expenditures (PCE) Index--See Table 2.

Explanation: The CPI does not include government purchases and investment goods and is the most widely used price index for consumer spending. The PCE index is an alternative that can also be used.

Table 2: Price Index Series (last updated April 2010)

^1. http://www.bea.gov/national/nipaweb/TableView.asp?SelectedTable=4&ViewSeries=NO&Java=no&Request3Place=
N&3Place=N&FromView=YES&Freq=Year&FirstYear=1977&LastYear=2009&3Place=N&Update=Update&JavaBox=no

^2. CPI Index-All Urban Consumers, Bureau of Labor Statistics (http://data.bls.gov/cgi-bin/surveymost?cu) (U.S. All items, 1982-84=100)

^3. http://www.bea.gov/national/nipaweb/TableView.asp?SelectedTable=73&ViewSeries=NO&Java=no&Request3Place=N&3Place=
N&FromView=YES&Freq=Year&FirstYear=1977&LastYear=2008&3Place=N&Update=Update&JavaBox=no

2. Pooling Two or More Years of MEPS Expenditures For Average Annual Estimates - Occasionally, sample size limitations make it necessary to pool two or more years of MEPS expenditure data to produce expenditure estimates of acceptable precision (for example, to analyze infrequently occurring health conditions or small demographic subpopulations).

• Guideline 1 (total): If pooling total expenditures, use the Personal Health Care Expenditure (PHCE) Price Index, CMS Office of the Actuary (see Table 3, column 1).
• Guideline 2 (service type): If pooling only one type of health care expenditure, for example prescription drugs, then use the price index for that component of the PHCE (see Table 3, columns 2-10) .

Explanation: Pooling expenditures requires a price index that is specific to health care services. Essentially, we are trying to mimic an annual expenditure estimate rather than compare inflation adjusted resources used for health care over time. But health care prices change from year to year, and faster than overall price inflation, so we need to account for these health care price changes when pooling.

The PHCE is recommended over two alternatives, the CPI component specific to medical care (CPI-M) and the GDP price index for medical care. The PHCE reflects total personal health care expenses, which is more appropriate than the CPI-M which only reflects out of pocket expenses. The PHCE is preferable to the GDP price index for medical care because the GDP index includes some expenses extraneous to MEPS such as those for medical research and public health programs. The CMS Office of the Actuary constructs the PHCE based on components of the CPI-M and the Producer Price Index (PPI) for hospital care (see Table 3).

Table 3: PHCE and Component Price Indices - Part 1 of 2

Table 3: PHCE and Component Price Indices - Part 2 of 2

^1. Indices for 1977 and 1987 are also included in Table 2 to facilitate adjustments to estimates from the National Medical Care Expenditure Survey (NMCES) and National Medical Expenditure Survey (NMES), respectively.

^2. Nursing home care component of overall index not included as separate column in table because MEPS does not cover residents of institutions.
Contact Office of Actuary at Center for Medicare and Medicaid services or go to http://www.cms.gov/NationalHealthExpendData/downloads/dsm-09.pdf for more information on PHCE.

^3. http://data.bls.gov/cgi-bin/dsrv?pc (search text for hospitals)

^4. http://data.bls.gov/cgi-bin/dsrv?cu (Not seasonally adjusted/U.S. city average/Current/search text for item)

^5. Category discontinued in 2010. Go to http://www.bls.gov/cpi/cpichg2010.htm for more information.

3. Analyses Using Annual Income Data

Individual and family income data collected in MEPS are commonly used as covariates in behavioral analyses of health care spending and/or to construct measures of burden (e.g. ratio of out-of-pocket medical expenses to income). The following general guideline for adjusting income data mirrors the recommendation in section 1.2 above for adjusting out of pocket expenditures.

• Guideline: Use the Consumer Price Index (CPI all-item all urban consumers) or alternatively the Personal Consumption Expenditures (PCE) Index--See Table 2.