Congressional Budget Office

Presentation to the National Center for Health Statistics

Peter OrszagDirector

August 11, 2008

Federal Spending Under CBO’s AlternativeFiscal Scenario

Percentage of Gross Domestic Product

1962 1972 1982 1992 2002 2012 2022 2032 2042 2052 2062 2072 2082

0

10

20

30

40

Medicare and Medicaid

Actual Projected

Social Security

Other Spending (Excluding debt service)

Misdiagnosing the Problem

Most discussions in media:

aging and demographics

Most of the fiscal problem:

rising cost per beneficiary

(not number or type of beneficiaries)

Sources of Growth in Projected Federal Spending on Medicare and Medicaid

Percentage of GDP

2007 2022 2037 2052 2067 2082

0

5

10

15

20

Effect of Aging Alone

Interaction

Effect of Excess CostGrowth Alone

Before we all get too depressed…

Embedded in the nation’s centrallong-term fiscal challenge appears to be a substantial opportunity:

Can we reduce health carecosts without impairing healthoutcomes?

Medicare Spending per Beneficiary in the United States, by Hospital Referral Region, 2005

Variations Among Academic Medical Centers

UCLA Medical Center

Massachusetts General Hospital

Mayo Clinic(St. Mary’s Hospital)

Biologically Targeted Interventions: Acute Inpatient Care

CMS composite quality score 81.5 85.9 90.4

Care Delivery―and Spending―Among Medicare Patients in Last Six Months of Life

Total Medicare spending 50,522 40,181 26,330

Hospital days 19.2 17.7 12.9

Physician visits 52.1 42.2 23.9

Ratio, medical specialist / primary care 2.9 1.0 1.1

Use of Biologically Targeted Interventions and Care-Delivery Methods Among Three of U.S. News and World Report’s “Honor Roll” AMCs

Source: Elliot Fisher, Dartmouth Medical School.

What Additional Services Are Provided in High-Spending Regions?

Source: Elliot Fisher, Dartmouth Medical School.

CBO’s Activities in Analyzing Health Care

New Hires and Expanded Staffing– New deputy assistant director (Keith Fontenot) in the Budget

Analysis Division– Increase in health staff agency-wide from 30 FTEs to

50 FTEs – FY 2009 plans

Reports and Analysis in 2008– Critical Topics in Health Reform– Health Options

Examples of CBO’s Uses of NCHS Data Sets

Comparisons of measured BMI and self-reported BMI (NHANES)

Estimates of Medicaid long term care costs (National Nursing Home Survey)

Analysis of the impact of increases in certain chronic conditions on the receipt of SSDI (NHIS)

Current CBO Projects Using NCHS Data Sets Linked to Administrative Data

Estimating the effects of current and past obesity on Medicare spending– NHANES I Epidemiologic Follow-up Study (20-year follow-

up) linked to Medicare claims data

Comparing Medicare spending for individuals with and without health insurance prior to becoming eligible for Medicare– NHIS data linked to Medicare claims data

CBO’s Wish List for Data

Richer Mortality and Longitudinal Data Data on Social Environment Development of a Single National

Health Indicator

Mortality: A Key Health Outcome

Used to measure national health trends Used to measure disparities by race, ethnic origin Used to assess differences in quality across

providers Trend toward looking for value and efficiency in

health care is generating interest in more refined measures of health outcomes, but mortality will always be key indicator

Increase in Life Expectancy, and Increase in Difference in Life Expectancy by Economic Status

Source: Data from Singh and Siahpush (2006) and CDC.

Years

At Birth At Age 65

0

1

2

3

4

Increase in Average Life Expectancy, 1980–2000

Increase in Difference in Average Life ExpectancyBetween Lowest and Highest Decile, 1980–2000

Mortality Data

Richer data sought linking comorbidities, educational attainment, other demographic considerations

– Periodic efforts to quantify the level of known problems via vehicles like the National Mortality Follow Back survey would be very useful

– Including/improving SES information in mortality data would be valuable

– Opportunities to include and/or improve information on nation of origin and duration of domestic residence would enrich the data

Longitudinal Data

Cross-sectional data are insufficient when the lag between treatment and outcome is long

Collection of longitudinal data with a sufficient sample size to analyze multiple interventions/outcomes is costly

Opportunities to enhance utility of cross-sectional data (following subsamples, linking with administrative data) offer an alternative approach– For example, tracking NHANES subpopulations with similar

conditions, but different medical interventions

Social Environment and Impact on Disease

The importance of social integration is well known– Better mental health– Lower heart disease, mortality risk

Social environment affects health behaviors– Diet, physical activity, smoking

Social environment affects perceptions of health– Before unification, low self-reported back pain in East

Germany– After unification, reports of back pain increased in East

Germany, possibly because of exposure to West German media reports on the topic

Existing Data on Social Environment

NHANES– Sources of social and financial support– Church attendance– Number of friends

Hispanic Community Health Survey– Acculturation

Framingham Heart Study– Participants listed friends/family contacts for follow-up– Overlap between participant and contacts allowed

researchers to analyze impact of social v. geographic closeness on changes in obesity, smoking status

Future Data on Social Environment

National Longitudinal Study of Adolescent Health– Includes questionnaires for parents, siblings, peers, and

school administrators and interviews with romantic partners

Proposed Community HANES– Similar to NHANES, but focused in a few small areas– Would include data on physical environment; could include

data on social network

Single National Health Quality Indicator

Mortality/life expectancy is often used as a proxy for average health of the population

Single indicator is convenient for tracking progress over time, but mortality may not be the best single measure

A new national indicator could incorporate a more comprehensive range of health measures