Transcript Disease Based Price Indexes

U.S. BLS Plans for Developing
Disease Based Price Indexes
Michael W. Horrigan
Associate Commissioner
May 10th 2010
Presentation Outline
Motivation
 Recommendations for producing
disease based price indexes
 Issues and challenges in estimating
disease based price indexes
 CPI approach
 PPI approach
 Concluding remarks

2
MOTIVATION
National Health Concern
National Health Expenditures
As a Percent of GDP and PCE
25.0%
15.0%
10.0%
5.0%
0.0%
1960
1962
1964
1966
1968
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
Percentage
20.0%
Year
Ratio to GDP
Ratio to PCE
4
Health Inflation Compared to
Overall Inflation
All Items CPI v. Medical CPI
1960-2008
18
16
14
12
10
8
6
4
2
0
All Items CPI
Medical CPI
5
Is this price growth
alarming?
For over forty years, many claim that the CPI
Indexes are upwardly biased.
 Many believe that bias occurs because we
use the wrong concept.
 We price health care services and goods,
and not the treatment of the disease.

6
Disease based concepts
for Medical Expenditure





(Triplett, 1999) Patients use medical services for
human repair. Consumers use auto body shops for
car repair.
For a car repair, the consumer pays one price, and
the shop buys all the parts and labor to fix the car.
Patients do not pay one provider one price for a
human repair.
They separately purchase physician visits, RX, etc.
There is no market price for the entire treatment of
diabetes, as there is for a car repair.
Triplett suggests reporting by disease and not
service.
7
Examples of potential upward bias in
CPI medical indexes

Shift in the treatment of a cataract disease
from in-hospital to out-patient facility care
reducing the price of treating the disease –
Shapiro and Wilcox (2001).

Substitution from office visits to the use of
pharmaceutical medications reducing the cost
of treating mental illness - Berndt et al
(2000).
8
Major Medical Innovations
Affecting Input Use
Innovation
Service/Product
Disease Treated
MRI and CT
Diagnostic/Phys.
Visits
various
ACE inhibitors
RX/Hospitals
hypertension
Balloon angioplasty
Surgery/Hospitals
coronary artery disease
Statins
RX/Hospitals
high blood cholesterol
Mammography
Diagnostic
breast cancer
Coronary Bypass
Surgery
coronary artery disease
H2 blockers
RX
stomach
SSRI anti-depressants
RX/Therapy Visits
depression
Cataract extraction
Inpatient/Outpatient
cataracts
Hip and knee replacement
Inpatient/Outpatient
orthopedic
Biopharmaceuticals
RX/Hospitals/Phys.
mostly cancer
RECOMMENDATIONS FOR
PRODUCING DISEASE BASED
PRICE INDEXES
Recommendations

Triplett (1999) proposed that BEA and the NHEA
report medical expenditures by disease.

At What Price (2002), Recommendation 6-1,
established a methodology for measuring disease
based price indexes.
 Use “claims database to identify and quantify the inputs
used in” the treatment of a disease.
 “On a monthly basis, the BLS should reprice” current medical
inputs keeping the quantities fixed.
 “every year or two” update the quantities of inputs used to
treat a disease.
 When updated, “the index will jump at the linkage points.”
11
Recommendations

CNSTAT panel - Strategies for a BEA Satellite Health
Care Account (2009)
 Set up to review plans by the U.S. Bureau of Economic
Analysis to create a BEA satellite account that reports
medical PCE by disease.
 Very interested in U.S. BLS Producer Price Index Program to
produce disease based price indexes
12
ISSUES AND CHALLENGES IN
ESTIMATING DISEASE BASED
PRICE INDEXES
Issues and challenges

An episode of treatment often, if not
always, cross provider classes
Treatment protocol and reimbursement
requirements
– Eg., knee replacement surgery
Actual path of discovery and treatment can
vary widely
14
Issues and challenges

Episode of treatments is not a concept for
which universe frames exist for drawing
stratified probability samples based on
revenue.

Second best alternative is to aggregate
across provider classes by mapping coding
structures for each provider to diseases

Challenge and growing problem of comorbidities
15
Issues and challenges

Protocols for treating disease can
change over time
Changes in protocols can occur within and
across provider classes
Some changes in protocols may represent
a quality change of the same protocol
16
Issues and challenges

Deciding when a substitution has taken
place is difficult
Independent medical expertise
Comparativeness effectiveness research
Penetration rate of substitute protocols
Need to continue pricing the old and new
treatment protocols after deciding a
substitution has taken place?
17
CPI APPROACH
Using the Medical
Expenditure Panel Survey (MEPS)
MEPS does a survey for medical expenditures and
medical utilization. Substitutions toward less costly
inputs should be in the data.
 It is representative of the civilian non institutionalized
population.
 It surveys both households and providers on the
disease contracted and the use of goods and services
to treat those diseases.
 There is no charge for use.

19
Using MEPS to Generate Price Indexes
Organized by Disease




Merge the MEPS Conditions file and Event Files
For each disease get per patient input quantities.
Use CPI monthly relatives for physicians, RX,
and hospitals to measure monthly price growth.
Each year the quantity of the inputs used for
each disease are updated. If there is a
substitution to a less expensive input, there will
be a downward jump in the index.
20
Average per Patient Quantities(Utilizations)
1998-2004
21
Mental Disorders Example
Year
1998
1999
2000
2001
2002
2003
2004
Mean
Mean
Mean
Hospital Emergency
Mean
Office
Hospital
Number
Room
Outpatient
Visit Mean Admissions of Nights
Visits
Visits
4.73
4.39
4.50
3.79
3.88
3.83
3.99
0.05
0.05
0.03
0.05
0.04
0.04
0.03
0.95
0.52
0.29
0.41
0.39
0.42
0.40
0.03
0.03
0.03
0.05
0.04
0.04
0.04
0.44
0.46
0.33
0.18
0.18
0.21
0.22
Mean
Prescriptions
1.86
1.98
2.02
2.03
2.01
2.02
2.06
1999-2004 Cummulative Index Growth for Mental Disorders
Accounting for less expensive inputs
Fixed 1998 Inputs
7.20%
35.50%
22
Results
Expenditure
Method
Fixed
Quantities
Quantities
Updated Yearly
Adjusted for
Comorbidities
Total
Expenditures
35.85%
33.2%
30.91%
Out of Pocket
Expenditures
28.31%
31.63%
30.57%
BLS CPIScope
30.32%
30.55%
28.81%
23
Why Do the Results differ by
Expenditure Method?
Most of the savings accruing from the shift from inpatient to
outpatient hospital goes to third party payments.
Hospital prices are rising more rapidly than physician or
pharmaceutical prices, and an index based on total
expenditures has a higher hospital weight than an index
based on out of pocket payments. Therefore both indexes
for out of pocket payments are less than the index for
total expenditures.
Consumers pay a very small fraction of total inpatient
expenses. A shift from inpatient to outpatient for a
disease category increases the share of out of pocket
expenditures for treating that particular disease category.
24
Accounting for utilization changes
decreases the price index for
Diseases of the respiratory system
Other conditions
Diseases of the circulatory system
Diseases of the genitourinary system
Neoplasms
Diseases of the musculoskeletal system and connective tissue
Diseases of the digestive system
Diseases of the blood and blood-forming organs
Complications of pregnancy, childbirth, and the puerperium
Mental disorders
25
Accounting for utilization changes
increases the price index for
Diseases of the skin and subcutaneous tissue
Infectious and parasitic diseases
Congenital anomalies
Injury and poisoning
Diseases of the nervous system and sense organs
Endocrine, nutritional, immunity disorders (includes
diabetes and high cholesterol)
26
Key Findings
Increased utilization intensity for diabetes, and
cholesterol management.
 Largest index drop in Mental Disorders.
 The savings to third party payments differs
from out of pocket savings.
 The prices for inpatient hospital services are
rising faster than for other services. This has a
greater impact on third party payments than
out of pocket payments.

27
PPI APPROACH
PPI Future plans
U.S. Census Bureau is collecting revenue by
chapter titles of the ICD-9 manual.
 BLS can aggregate items from each of the
following industries to these same disease
categories:

Hospitals (DRGs)
Offices of Physicians (ICD-9, CPT)
Diagnostic Imaging Centers (ICD-9)
Medical Laboratories (ICD-9)
Pharmacies (Primary Therapeutic Equivalent)
29
PPI Future Plans

When does a substitution take place?
Need for independent judgment on
changes in what is a change in the
standard treatment protocol.
Plans to continue pricing each item for
current medical price indexes even after a
directed substitution has taken place and a
price drop is recorded.
30
PPI Future Plans

Quality change
CMS data on adherence by hospitals on
adherence to standard protocols for heart
attack, heart failure, and pneumonia.

Resampling and changes in weights
(utilization rates)
Laspeyres fixed quantity formula
Resampling is done every 7 years.
31
CPI and PPI Future Plans
CPI to update the results of their MEPS
based research every year and publish
it in a working paper on the BLS web.
 PPI will get the Census weights in 2010
and begin calculating disease based
price indexes in 2011.

Use of MEPS data also being considered
32
Disease based price indexes
and health care reform

CPIs important in measuring whether or
not health care reform lowered the out
of pocket costs to consumers.

PPIs important in measuring the total
costs of health care reform no matter
the source of financing.
33
Contact Information
Michael Horrigan
Associate Commissioner
Office of Prices and Living
Conditions
202-691-6960
[email protected]
The CNSTAT Formulae
Pdit = Price of input service i used to treat disease d in period t.
Qdir = Quantity of input i used to treat disease d in period r.
When quantities are not updated:
P

P
dit
I dt
Qdir
i
Qdir
dit 1
i
When quantities are updated:
P

P
dit
I dt
Qdir 1
i
Qdir
dit 1
i
P

P
Qdir 1
dit 1
i
dit
Qdir 1
i
Qdir
dit 1
i
P

P
Qdir 1
dit 1
i
 Jump from Q change  Price change only
35
Example - Mental Illness
Price of Office Visit = $200 in period 1; = $220 in period 2.
Price of RX
= $30 in period 1; = $33 in period 2.
The price of all inputs increase 10%.
Quantity of Office Visits = 4 in period 1 ; = 1 in period 2.
Quantity of RX
= 0 in period 1 ; = 4 in period 2.
The price of all inputs are up 10%.
Disease Based Index:
Pdoc ,2Qdoc ,2  PRX ,2QRX ,2
Pdoc ,1Qdoc ,1  PRX ,1QRX ,1

Pdoc ,1Qdoc ,2  PRX ,1QRX ,2
Pdoc,1Qdoc,1  PRX ,1QRX ,1

Pdoc ,2Qdoc ,2  PRX ,2QRX ,2
Pdoc,1Qdoc,2  PRX ,1QRX ,2
220*1  33* 4 200*1  30* 4 220*1  33* 4


200* 4  30*0 200* 4  30*0 200*1  30 * 4
.44

.4

1.10
Index
= Input Effect  Price Effect
36
36