Transcript Introducing the New BEA Health Care Satellite Account

Introducing the New BEA Health
Care Satellite Account
Abe Dunn, Lindsey Rittmueller, and Bryn Whitmire
SEM Conference, Paris
24 July 2015
Health care is a large and increasing share of GDP
$18
Health and non-Health Related Expenditures of GDP
$16
17.4%
$14
Expenditures (trillions)
$12
$10
$8
$6
$4
$2
GDP without health expenditures
bea.gov
2012
2010
2008
2006
2004
2002
2000
1998
1996
1994
1992
1990
1988
1986
1984
1982
1980
1978
1976
1974
1972
1970
1968
1966
1964
1962
1960
$-
National Health Expenditures
2
Health Care Satellite Account (HCSA)
 The goal of the satellite account is to improve
our understanding of health care spending in
the U.S. by developing disease-based statistics
 Development advocated by many health economic experts
 Scitovsky (1964) “An Index of the Cost of Medical Care – A
Proposed New Approach”
 Cutler, McClellan, Newhouse and Remler (1998) “Are Medical
Prices Declining? Evidence from Heart Attack Treatments”
 Frank, Berndt, Busch (1999) “Price Indexes for the Treatment of
Depression”
 Two reports of the National Academy of Sciences’ Committee
on National Statistics
bea.gov
3
Contribution of our work is to redefine the
output of the health care sector
For example
 Output = number of patients treated for breast cancer
 Expenditures = spending on the treatment of breast cancer
 Price = average spending per treated patient for breast cancer
bea.gov
4
Implication 1: Health care spending will be reported by
disease classes
bea.gov
5
Implication 2: Redefining output also implies new
price indexes
 The new price indexes
are the change in average
expenditure per episode
for each disease
bea.gov
6
Implication 2: Redefining output also implies new
price indexes
 The new price indexes
are the change in average
expenditure per episode
for each disease
 Disease-based indexes can rise
slower than traditional service
price indexes with shifts in
treatments
bea.gov
7
Implication 2: Redefining output also implies new
price indexes
 The new price indexes
are the change in average
expenditure per episode
for each disease
 Disease-based indexes can rise
slower than traditional service
price indexes with shifts in
treatments
 Disease-based indexes can also
rise faster than traditional
service price indexes
bea.gov
8
HCSA presents two alternatives covering the 2000-2012 period
1. “MEPS Account” – using Medical Expenditure Panel Survey
(MEPS)
2. “Blended Account” – MEPS, MarketScan® claims data, and
Medicare claims data
www.bea.gov/scb/index.htm
bea.gov
9
Key points
 The Blended Account produces more stable and less
volatile disease-based statistics
bea.gov
10
“Big Data” Motivation – Disease-Based Price Indexes
Nervous System
Musculoskeletal
130
120
110
100
90
80
70
60
50
MEPS
bea.gov
Blended
MEPS
Blended
11
Key points
 The Blended Account produces more stable and less
volatile disease-based statistics
 Improves our understanding of health care spending
trends
bea.gov
12
Key points
 The Blended Account produces more stable and less
volatile disease-based statistics
 Improves our understanding of health care spending
trends
 Disease-based indexes better reflect the pricing of health
care treatments and will allow for evaluation of quality
change
bea.gov
13
Organization
 Methods
 Data
 Results
 Next steps
bea.gov
14
Allocating expenditures across disease
categories
 Many potential strategies for disease allocation
 No consensus on “best” method
 Currently applies a simple method with widespread use:
 Apply a primary diagnosis approach using the Clinical
Classification Software (CCS) (263 categories)
 Uses “person-based” approach when primary diagnosis
approach is not possible
bea.gov
15
Medical Care Expenditure Indexes (MCEs)
𝑀𝐶𝐸𝑑 =
𝑐𝑑,𝑡
𝑐𝑑,0
𝑐𝑑,𝑡 = Average expenditure per treated patient at time, t, for
condition, d . (𝑡 = 0 is the base period.)
 MCE indexes constructed at disaggregate level
 Report by disease category and in the aggregate
 Nominal expenditure
 Price index (MCE)
 Real expenditure
bea.gov
16
MEPS Data
 Nationally representative sample with approximately
30,000 individuals
 All sources of spending and associated CCS diagnosis
codes
 Raw MEPS data files used (pool 2 years of data)
bea.gov
17
MarketScan® and Medicare Claims Data
 Commercially-insured patients from the MarketScan®
Data from Truven Health
 More than 2 million enrollees in each year
 Convenience sample  application of population weights
 Medicare patients from 5 percent random sample of
enrollees
 Approximately 2 million enrollees each year
 Prescription drug expenditures per episode imputed using
MEPS for 2000-2012 sample period
 Excludes Medicare Advantage enrollees  application of
population weights
bea.gov
18
Construction of Blended Account
 Use population weights from MEPS to fold in
data from different sources
MEPS Other
(e.g. Uninsured,
Medicaid)
Medicare
Population 
Medicare FFS
5% Sample
bea.gov
Privately
Insured 
MarketScan®
19
Construction of Blended Index
 After claims data is incorporated, the Blended Account is
constructed identically to the MEPS Account
 Expenditures and MCE indexes are computed for each of
the 263 CCS categories
 Report 18 Disease Categories (CCS Chapters) and
aggregate
bea.gov
20
MCE MEPS, and MCE Blended yields
faster growth than PCE (2009=100)
110
Health Price Index for the
PCE, MEPS, and Blended Accounts
105
100
95
Price (2009=100)
90
PCE Health
85
MCE MEPS
80
MCE Blended
75
70
65
60
2000
bea.gov
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
21
Effects on Annual Real Growth Rates
*Based on 2000-2010 Estimates
PCE Health by Function
Actual PCE Health
Using MCE MEPS
Using MCE Blended
3.3%
2.4%
2.0%
Overall PCE
Actual PCE
Using MCE MEPS
Using MCE Blended
2.1%
1.9%
1.8%
GDP
Actual GDP
Using MCE MEPS
Using MCE Blended
bea.gov
1.6%
1.5%
1.5%
22
Total Spending by Disease Category
$275
Symptoms
Circulatory
$225
Expenditures (billions)
Musculoskeletal
$175
Respiratory
Endocrine
Nervous
Neoplasms
$125
$75
$25
2000
bea.gov
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
23
Trends in disease-based price indexes are volatile
using the MEPS index
130
120
Price Index (2009=100)
110
100
90
Symptoms
Circulatory
80
Musculoskeletal
Respiratory
Endocrine
70
Nervous System
Neoplasms
60
2000
bea.gov
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
24
Trends in disease-based price indexes are volatile
using the MEPS index
130
120
Price Index (2009=100)
110
100
90
Symptoms
Circulatory
80
Musculoskeletal
Respiratory
Endocrine
70
Nervous System
Neoplasms
60
2000
bea.gov
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
25
Increases in cost per case contributed 64 percent
to per capita spending growth
36%
64%
bea.gov
26
Decomposition of the Contributions to
Annual Spending Growth
10%
9%
8%
Annual Growth (percent)
7%
6%
5%
4%
3%
2%
1%
0%
2001
2002
2003
2004
2005
disease-based price growth
bea.gov
2006
2007
prevalence growth
2008
2009
2010
2011
2012
per capita growth
27
Decomposition of the Contributions to
Per Capita Spending Growth for 2000-2012
4.4
1.1
Medical services by disease
5.0
2.7
3.5
Symptoms
0.1 Circulatory
4.7
-0.5
1.9
4.2
Musculoskeletal
Respiratory
3.6
3.5
5.3
0.6
4.6
-0.6
0.5
4.8
Nervous
Neoplasms
Injury and poisoning
4.3
-0.3
Endocrine
1.1
4.9
Genitourinary
Digestive
3.1
2.0
Mental illness
6.1
2.1
4.7
0.5
4.1
Skin
0.0Pregnancy
7.2
-1.0
0.0
1.0
2.0
Infectious
3.0
1.1
4.0
5.0
6.0
7.0
Other
8.0
9.0
10.0
Annual Growth (percent)
Disease-Based Price Index
bea.gov
Treated Prevalence
28
Per Capita Spending Growth for Select
CCS Categories
2.75
Coronary Atherosclerosis
Hypertension
2.50
2.25
2.00
1.75
Per Capita
1.50
Per Case
1.25
Prevalence
1.00
0.75
2.75
High Cholesterol
2.5
Diabetes no complications
2.25
2
1.75
1.5
1.25
1
0.75
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
bea.gov
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
29
Conclusion
 Improves our understanding of spending trends
 Another step in broader goal to improve health statistics
 Another step in broader goal to improve health statistics
 A lot more work to be done…
bea.gov
30
Next steps
 Representativeness (e.g., Medicare Advantage and
Commercial HMO enrollees)
 Expenditure allocation
 Coverage over time
 Quality adjustment
bea.gov
31
www.bea.gov/national/health_care_satellite_account.htm
bea.gov
32
EXTRA SLIDES
bea.gov
33
Recent slow down in the annual growth in
nominal health spending (2001-2013)
10%
PCE Health
8%
NHEA (CMS)
6%
4%
2%
0%
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
bea.gov
34
Differences between an MCE and PPI
Examples: (Ceteris Paribus)
MCE PPI
Change from high cost inpatient visit to lower cost outpatient visit
↓
-
Higher intensity procedures used in physician offices
↑
-
Change from restrictive insurance plan to generous plan
↑
-
Higher prices for physician office procedures
↑
bea.gov
↑
35
Disease allocation methods
General strategies for disease allocation:
1.
2.
3.
Encounter-based approach (e.g., primary diagnosis method)
Episode-based approach
Person-based approach
 BEA has not settled on a “best” method
 Apply a primary diagnosis approach using the CCS
classification system (263 Clinical Classification Software
(CCS) categories)
 Simplicity and widespread use
bea.gov
36
bea.gov
37
Total Spending by Disease category
(Using Blended method for 2010)
Diseases of the circulatory system
Symptoms; signs; and ill-defined conditions
Diseases of the musculoskeletal system and…
Top 5 categories
account for 51
percent of spending
Diseases of the respiratory system
Endocrine; nutritional; and metabolic diseases and…
Diseases of the nervous system and sense organs
Neoplasms
Diseases of the genitourinary system
Injury and poisoning
Diseases of the digestive system
Mental illness
Infectious and parasitic diseases
Diseases of the skin and subcutaneous organs
Complications of pregnancy; childbirth; and the…
Residual codes; unclassified; all E codes
Diseases of the blood and blood-forming organs
Congenital anomalies
Certain conditions originating in the perinatal period
bea.gov
0
50
100
150
200
250 38
Increases in cost per case contributed 73 percent
to per capita spending growth?
2.2
Health Care
Spending
2
2.2
2
1.8
Real per
capita
spending
1.8
71%
1.6
1.6
1.4
1.4
Population
Growth
1.2
1
1
0.8
0.8
bea.gov
Real MCE price
index
31%
PCE Prices
1.2
MCE price
index
PCE prices
39
Effect of HMO Inclusion - MarketScan
MCE Comparison - HMOs Included - MarketScan ETG
1.4
1.35
1.3
1.25
1.2
1.15
1.1
1.05
1
0.95
0.9
2003
bea.gov
2004
2005
2006
2007
Combined HMO & Non-HMO MCE
2008
2009
Non-HMO MCE
2010
40
Effects of Severity Adjustment
MCE Comparison - MarketScan ETG
MCE Comparison - Medicare CCS
1.60
1.60
1.50
1.50
1.40
1.40
1.30
1.30
1.20
1.20
1.10
1.10
1.00
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Non-Severity
Adjusted
bea.gov
ETG Grouper Severity
Adjusted
Regression Severity
Adjusted
1.00
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Non-Severity
Adjusted
Regression Severity
Adjusted
41
Coverage over time
 Historic estimates
 Aizcorbe and Highfill (2015) – back to 1980
 Timely estimates
 BLS work – Bradley (2014)
bea.gov
42
Recent research shows growth in a disease-based price
index can rise faster or slower than the PCE price index
10.0%
9.0%
8.0%
Disease-based index
7.0%
PCE Health
6.0%
5.0%
4.0%
3.0%
2.0%
1.0%
0.0%
1980-1987
bea.gov
1987-1997
1997-2006
Source: Aizcorbe and Highfill, “Medical Care Expenditure Indexes for the US, 1980-2006”
43
Quality Adjustment – Heart Attacks (AMI)
bea.gov
44
Schedule and Future Work
 Schedule
 SCB release of updated data for HCSA for 2011 and 2012 (later in 2015)
 Future work
 Continue to work with other agencies and health experts on consistent
measures of disease-based spending and prices
 Creating a longer time series and current estimates
 Evaluate the impact on Industry accounts
 Continue to evaluate data sources – MEPS, MarketScan®, Medicare,
Medicaid, nursing homes, and others
 Evaluate impact of severity
 Evaluate quality adjustment
bea.gov
45
Implication for Industry Accounts: New Price Indexes
imply change in Value Added and Gross Output
 Real Value Added = Gross Output – Intermediate inputs
 Input prices will stay the same for intermediate
commodities
 Use new disease-based price index for all affected
disaggregated sectors
 Our current method in the Health Care Satellite Account
allocates spending proportionally across all industries
 Productivity in certain health sectors will change
 More research is needed to evaluate other methods to
incorporate these new indexes
bea.gov
46
Industry Example: Average Annual Real Growth Rates
between 2000 and 2010 for Gross Output and Value Added
Gross Output
Current
Industry Code
Industry Description
Value Added
Alternate
Blended
Current
MEPS
Alternate
Blended
MEPS
Average Annual Quantity Growth, 2000-2010
62
Health care and social assistance
3.4%
2.1%
2.3%
2.8%
0.7%
1.1%
621
Ambulatory health care services
3.4%
1.9%
2.2%
3.4%
1.0%
1.5%
622
Hospitals
3.6%
1.7%
2.1%
2.6%
-0.6%
0.1%
Average Annual Price Growth, 2000-2010
62
Health care and social assistance
2.8%
4.1%
3.9%
3.2%
5.4%
4.9%
621
Ambulatory health care services
2.4%
3.9%
3.6%
2.5%
4.9%
4.5%
622
Hospitals
3.2%
5.0%
4.7%
3.9%
7.2%
6.6%
bea.gov
47