Transcript Bruce Guthrie Glenna Auerback Andrew

What changed when incentives
changed in California Medicaid?
Bruce Guthrie
Glenna Auerback
Andrew Bindman
Pay-for-performance as panacea?
“Pay for performance’s goal is not simply to
reward those who perform well or to reduce
costs. Rather, it is a mechanism to align
incentives to encourage ongoing
improvement in a way that will ensure highquality care for all.”
Institute of Medicine 2006
Pay-for-performance in Medicaid
• Public insurance for the poor & disabled
• Federally defined minimal provision, but
considerable State autonomy
• Mixed public-private provision
• Cash-strapped
California Medicaid
Medi-Cal Managed Care
• 8 counties with a single
health plan
• 14 counties with a
choice of health plan
The ‘pay’ in Medi-Cal pay-for-performance
• Auto-assignment of new enrollees who
don’t choose a plan
– ~25% enrollee turnover annually
– ~25% of new enrollees don’t choose a plan
– ~5% of plan membership at risk each year
• Attractions
– Cost neutral and simple to implement
• Disadvantages
– Variable and opaque incentive
The ‘performance’ in Medi-Cal
pay-for-performance
• Composite quality score determines autoassignment share
• Five HEDIS measures
– Childhood immunisations
– Well child checks
– Adolescent well care
– Timeliness of pre-natal care
– Appropriate medications for people with asthma
Research questions
Qualitative
1. Were plans incentvized?
2. What did plans do in response?
3. What are the perceived consequences
Quantitative
4. Did incentivized quality change?
5. Did non-incentivized quality change?
Methods - qualitative
• Documentary analysis
– Public reports
– Advisory Group minutes and briefing notes
• Semi-structured interviews with:
– Plan CEOs, Medical Directors, QI Directors
– Other members of Stakeholder Advisory
Group
– 20 interviews with 29 participants
– 12 out of 15 plans in affected counties
Methods - quantitative
• Comparison of changes in quality in:
– Managed care counties with choice of plan (intervention)
– Managed care counties with a single plan (control)
• Difference-in-differences analysis of changes in
quality from ‘before’ to ‘after’ implementation
• HEDIS data for 4 incentivized and 4 nonincentivized measures 2004-2007
• Preliminary/premature results for discussion and to
demonstrate methods
Q1. Were plans incentivized?
• Performance based auto-assignment is an
incentive
– Members as money
– Members as mission
• But it’s one incentive among many
– State regulation
– Internal motivation to deliver high quality
– Business case
“We’re doing the Lord’s work, we’re
protecting the safety net.”
Chief Executive Officer
“I think of it as membership, but of course,
marketing and finance think of it as dollars.”
Medical Director
“Well [auto-assignment] is definitely one of
the drivers, you know, of what are we going
to work on this year. … The other drivers,
you know, you’ve got the HEDIS, the
Minimum Performance Level drivers. You’ve
got your collaboratives.”
Medical Director
Q2. What did plans actually do?
• Member focused QI
– Information, reminders, incentives
• Provider focused QI
– Information, technical support, incentives
• Improve data collection
– Reliable data collection, data warehouses
• Change in focus more than de novo QI
“When they chose those five HEDIS rates,
those became the sacred five. … We have a
small provider incentive that is limited to a
certain number of providers … All the time,
people are asking me “Can we add another
one [provider]?” and Well Baby is not one of
the five… Would I rather spend that money
on one of the five, well yeah. Those are the
five.”
Quality Improvement Director
Q3. Perceived consequences
• Better HEDIS scores
– Better quality of care?
– Better quality of data?
• Risk of crowding out other QI activity
• Risk of decreased collaboration with
competitor plans in QI work with providers
“I think there is early evidence in increases
in our HEDIS scores that are having an
impact on patient care, but … there is a
reporting aspect to this as well. That you
could have an improvement in how you
collect data, that will also improve your
HEDIS scores.”
Medical Director
“I think what it’s done is made you have to
go out and spend a lot of money to try to
collect the data … So you’re actually kind of
diverting probably, dollars from providing
actual quality into documenting quality.”
Vice President
“I would prefer it not to be competitive. … I
think what’s most effective is change at the
provider level. And change at the provider
level requires co-operation among payers.”
Chief Executive Officer
Summary of qualitative findings
• Incremental not transformational
• Expect to improve incentivized measures
• Concern that non-incentivized care could
be made worse
• Concern about competition reducing
collaborative work with providers
Q4. Did incentivized quality change?
No difference (3 measures)
Childhood immunizations combo 2
Significant difference (1 measure)
Pay-for-performance plans
Pay-for-performance plans
Adolescent well care
Comparison plans
Comparison plans
100%
Percentage receiving appropriate care
Percentage receiving appropriate care
100%
80%
60%
40%
20%
0%
80%
60%
40%
20%
Imms2004
Imms2005
Imms2006
Year
Same for:
Timeliness of prenatal care
Appropriate asthma medications
AWC2004
AWC2005
Year
AWC2006
Q5. Did non-incentivized quality change?
No difference (3 measures)
Significant difference (1 measure)
Pay-for-performance plans
Pay-for-performance plans
Cervical cancer screening
Well Child checks 1st 15 months
Comparison plans
100%
Percentage receiving appropriate care
80%
Percentage screened
Comparison plans
100%
60%
40%
20%
80%
60%
40%
20%
0%
0%
CS2004
CS2005
Year
Same for:
Post-natal care
Chlamydia screening
CS2006
WC152004
WC152005
Year
WC152006
Implications for US policy
• More evidence that pay-for-performance
isn’t rapidly transformational
• Transparency of incentives
• Who should be incentivized
• Scope of pay-for-performance
• Competition with incentives vs a coherent
single system
Implications for the UK
• Pay for performance as a useful, but
uncertain tool
• Incentives for quality when cash is tight
• Pay for performance for UK hospitals?
• Policy debate about relative effectiveness of
competition vs collaboration vs command
– Competition between providers vs competition
between purchasers
Thank you!