Transcript National Health Accounts

The Value of Knowing:
National Health Accounts
0011 0010 1010 1101 0001 0100 1011
1
David M. Cutler, Harvard University
2
4
Questions We Need to Answer
0011 0010 1010 1101 0001 0100 1011
• US medical spending has doubled as a share
of GDP since 1975. Is it worth it?
1
2
• Where should we spend our research dollars
most productively?
4
• Which country is best at hockey?
What is Needed?
0011 0010 1010 1101 0001 0100 1011
• A way to tell what is happening in the
medical sector, and what it’s worth.
1
2
4
The Health Sector
0011 0010 1010 1101 0001 0100 1011
Inputs
• Medical spending
– By disease
• Behaviors
– Smoking, obesity
• Environment
• Genetics
Outputs
• Population health
– By demographic
group
1
2
4
The Analogy: GDP
0011 0010 1010 1101 0001 0100 1011
Inputs: Factor payments
• Wages paid to workers
• Raw materials
• Return to capital
Output: GDP
• Overall
• Consumption,
investment, and
government spending
• By industry
1
2
4
What We’re Up To
0011 0010 1010 1101 0001 0100 1011
1
Inputs
• Medical spending
• Behaviors
• Environment
• Genetics
2
3
Outputs
• Population health
1
2
4
The “industry” in medical
care is the disease
More on the Rationale
0011 0010 1010 1101 0001 0100 1011
“The BLS should develop a research
program to look beyond its current "market
basket" framework for the CPI…
“We strongly endorse a move in the CPI
away from the pricing of health care inputs
to an attempt to price medical care
outcomes.”
- The Boskin Commission
1
2
4
The Idea of a Satellite Account
0011 0010 1010 1101 0001 0100 1011
“[W]e recommend the development of satellite
accounts to report on selected activities not
included in the conventional accounts. Satellite
accounts can link to the existing economic
accounts as appropriate, but also expand into
areas that the NIPAs do not cover.”
- Beyond the Market
1
2
4
Medical Spending
0011 0010 1010 1101 0001 0100 1011
1
2
4
Medical Spending: Big Challenges
0011 0010 1010 1101 0001 0100 1011
• The industry
– MI vs. Diabetes
– Prevention vs. screening vs. treatment
• The data
1
2
– They don’t give conditions unless people say they are why they
sought care.
– We know what we spend on diabetes, but not what the average
diabetic spends.
• The level of analysis
4
– What disease to attribute to an office visit for a hypertensive
diabetic with a past MI?
Cost Model Approach
0011 0010 1010 1101 0001 0100 1011
Annual spending = condition categories + e
1
Note that this is a ‘person-based’ method
rather than an ‘encounter-based’ method.
2
4
Taking it Back Farther
0011 0010 1010 1101 0001 0100 1011
CHD
Smoking
Diabetes
BMI
Cancer
1
2
Spending
4
Population Health
0011 0010 1010 1101 0001 0100 1011
1
2
4
Components of Population Health
0011 0010 1010 1101 0001 0100 1011
• Mortality
• Quality of life
1
2
4
1. Official mortality data are
problematic
0011 0010 1010 1101 0001 0100 1011
• Accuracy of diagnosis coding is in doubt
• Doesn’t get at risk factors (obesity) or
precursor diseases (diabetes)
1
2
4
Mortality model approach
0011 0010 1010 1101 0001 0100 1011
• Estimating models of mortality, similar to
spending
1
Prob die = condition categories + e
2
4
• With these and the disease / risk factor
transition models, we will have a way to
determine ‘true’ cause of death
2. Quality of Life
0011 0010 1010 1101 0001 0100 1011
Health
Domain 1:
Domain 2:
Symptoms / Impairments
Symptoms / Impairments
Disease 1
Disease 2
1
2
Domain 3:
Symptoms / Impairments
4
Disease 3
Trend in Quality of Life, 1987-2004
0011 0010 1010 1101 0001 0100 1011
0.81
1
0.80
0.79
0.78
0.76
0.75
1987
2
4
Increased obesity
0.77
2001
2002
2003
2004
Ac
ti v
i ty
re
ss
iv
e
An
xi
ou
s
ft
al
ki
ng
ar
e
nd
/L
i
Be
W
lfc
Se
da
ry
ec
on
y
D
ep
l/S
ci
a
So
Pr
im
ar
Decrement on 0 to 1 scale
0.05
0.00
1
4
g
rin
ea
0.10
H
0.15
Vi
si
on
0011 0010 1010 1101 0001 0100 1011
Symptoms/impairments with
largest decrements
0.25
0.20
2
Disease Models
0011 0010 1010 1101 0001 0100 1011
1
2
4
A Catalog
0011 0010 1010 1101 0001 0100 1011
CHD
Stroke
Cancer
PNA
Depression
Huge improvements; medical care plays
big role (published in Health Affairs)
Large decline in mortality from stroke –
not clear why
Models have been developed; need to use
to answer this question
Large decline in mortality, likely related
to better medical care
No change in lifetime prevalence but big
reduction in current prevalence
1
2
4
What Will We Learn? My Guesses
0011 0010 1010 1101 0001 0100 1011
• By and large, technological advance has
been very important and cost effective
1
• In recent years, obesity trends have
significantly increased spending and
lowered health
2
4
– May have overwhelmed technology in parts