Medicare Spending, the Physician Workforce and Beneficiaries’ Workforce, and Beneficiaries Quality of Care

advertisement
Medicare Spending, the Physician
Workforce and Beneficiaries
Workforce,
Beneficiaries’
Quality
Qua
ty of
o Care
Ca e
Katherine Baicker
Harvard School of Public Health
Amitabh Chandra
Harvard Kennedyy School
1
High
g Costs – High
g Value?

Reform evaluated primarily on two criteria: coverage
and costs – but is cost right metric?



Low-value
Low
value spending has many negative consequences


Worthwhile reforms that improve value may or may not save
moneyy
Dangerous appeal of painless cost reduction
Health care dollars misallocated; slower wage growth; higher
uninsurance; cost of financingg public
p
insurance
International comparisons often used as evidence that
g
value
we could be ggettingg higher


Difficult inference problems
Mayy learn more from focusingg on domestic evidence
2
Qualityy Variation Even within Medicare
Source: Dartmouth Atlas of Health Care
3
Variation in Medicare Spending
p
g
Source: Dartmouth Atlas of Health Care
4
Motivation to Look at Relationship
p
Between Spending and Quality

Advantages of Medicare claims data




Relatively
R
l ti l complete
l t population
l ti
Relatively complete utilization
Path over time
But


May not generalize to those under 65
H d to get measures off quality
Hard
li and
d ill
illness that
h are not
contaminated by differences in practice pattern
5
Our Approach
pp

Following Jencks (2000; 2001), use QIO measures
aggregated to state level



6 common conditions: AMI,, breast cancer,, diabetes,,
heart failure, pneumonia, and stroke
Quality metrics that are less affected by underlying health
of patients – so not sensitive to risk adjustment
Regress on state-level Medicare spending (along
with
ith covariates)
ri t ) for
f r 2000-2001
2000 2001
6
Data

Medicare FFS Part A and B claims data



Aggregated to state level based on residence of enrollees
Adjusted for age
age-sex-race
sex race of enrollees
Adjusted for state-level price levels and inflation

Jencks:
k measures off quality
li

ARF



Physician workforce
Demographic covariates
MCBS: satisfaction and HMO penetration
7
Higher Spending
p
g Associated with Lower Quality
8
See Same Patterns in Components
p
9
See Same Patterns in Components
p
10
Higher
g
Spending
p
g Associated with
Lower Quality

Interpreting magnitude: in States spending $1,000
more per beneficiary



Overall quality rank was 10 spots lower
Of stroke p
patients,, 2.4%
% fewer prescribed
p
warfarin
Of female enrollees under 70, 2.1% fewer received a
mammogram within the last 2 years

Robust to inclusion of covariates (such as HMO
penetration
i and
d AMI iincidence
id
rate))

No difference in p
patient satisfaction
11
What Drives the Negative
g
Relationship?
p

Don’t have an experiment
p
((or quasi-experiment)
q
p
)
here


Can look at correlates for hints about things that may be
fruitful to explore
Hypothesis: the mix of the physician workforce
mayy affect both cost and quality
q
y

Note that this is a different question from the effect of
the size o
of the physician
physicia workforce
wo o ce – to answer
a swe our
ou
question we want to hold the size of the workforce
constant
12
Relationship
p between Composition
p
of
Workforce, Cost, and Quality
outcome  0  1%specialists  2 total MDs  covars+ 



Standard
d d llinear regression with
h severall covariates
Outcome is spending
p
g per
p beneficiaryy or quality
q
y
Can use different measures of specialist share, GPs,
RNs etc
RNs,
etc. as key regressor
13
Greater Specialization
p
of Workforce Associated
with Higher Spending, Lower Quality
14
Reverse for Higher
g
Share Primary/GP
y/
15
Suggestive
gg
– but Clearlyy not Definitive

May be that lack of coordination of care plays role




High intensity utilization may “crowd out” low intensity, highvalue care
S
Seems
associated
i d with
ihh
having
i specialists
i li comprise
i a greater
share of the physician workforce – consistent with other
evidence on “fragmentation”
g
Doesn’t mean specialists aren’t good at specialty care!
Further exploration into causal mechanisms
warranted

Alt
Alternative
ti stories,
t i pathways
p th
16
Implications
p

Need to think about underlying cause of disconnect
between costs and benefits

Financing and payment reform

V i i iin quality
Variation
li and
d cost




Both within and between local geographic areas (wide body of research
from Dartmouth and elsewhere))
Symptom of problem and evidence of inefficiency – not problem in and
of itself
Suggestive of potential for bundled payments,
payments promotion of
coordination
Focus on improving value,
value not just lowering cost

Will still have to make tough choices – will quickly run out of
qquality-improving,
y p
g, cost-reducingg improvements
p
17
Download