Dr. Ashish Jha's slides from the Center for Health Journalism webinar "Does ‘Pay for Performance’ Work?" 6.28.16
http://www.centerforhealthjournalism.org/content/does-pay-performance-work
4. +
Why is Pay for Performance attractive?
Has tremendous face validity
Works in other industries
Aligns incentives for better care:
Allows providers to do well when doing good
5. +
Incentives 1.0: What did we try?
Premier P4P
Began 2003
Small dollars
Process measures
8. +
And the news is discouraging too…
Headlines over the past 5 years:
“Health Affairs article finds Medicare’s pay-for-
performance did not spur quality improvement”
“New NEJM Report: Pay-for-performance…a bust”
“Paying doctors for quality doesn’t work”
“Medicare’s policy did not reduce infection rates”
9. +
What is the ACA doing for P4P?
A variety of new programs
Value-based purchasing
Hospital Readmissions Reduction Program
Hospital-Acquired Condition Reduction
Now: MACRA and MIPS
14. +
Which hospitals are getting penalized?
6.3% 5.7%
15.1%
20.9%
23.4%
17.8%
16.0%
7.2%
21.0%
30.4%
37.1%
27.0%
0%
5%
10%
15%
20%
25%
30%
35%
40%
Percentage Black Percentage
Hispanic
Divorced/never
married
Less than High
School Diploma
Lowest Quartile
of Household
Income
Medicaid
Enrollment
Low readmission rate hospital High readmission rate hospital
Barnett et al., JAMA IM 2015
15. +
ACA Reform #2: VBP (aka P4P)
Up to 2% of Medicare payments tied to:
Broad set of quality measures:
Processes
Outcomes
Patient Experience
Efficiency
17. +
Impact of VBP on Patient Experience
64
69
71
50
55
60
65
70
75
80
85
90
95
100
2008 2009 2010 2011 2012 2013 2014
PercentofPatientsRatingtheirHospital'9&10'
Pre-VBP Slope=
+1.46% per year
Post-VBP Slope=
+0.55% per year
Onset of VBP
18. +
What have we learned?
Incentives can move the needle
Simple measures
Narrowly focused
They can have unintended consequences
We need to understand the tradeoffs
Do they make care meaningfully better?
Jury remains out
19. +
Let’s reframe
Old question: “Does pay-for-performance work?”
New question: “How do we get pay-for-performance to
work?”
20. +
Incentives 2.0: What might it look like?
Bigger incentives?
Target a small number of outcomes?
Especially over the longer run
Across a broader set of measures
Structure it more simply
Play into intrinsic motivations
More nuanced approach to the safety net
This is an example of the variations in outcomes for acute myocardial infarction for Medicare patients entering a hospital in 2010. There are hospitals where outcomes are four times worse than those in other hospitals.
Similarly, for cancer care – we see large variations in cancer-related death rates across medical centers, across states, and across regions.
While the data are less well defined, there is plenty of reason to believe that there are clinically meaningful variations in use of evidence-based chemotherapy regimens, approaches to cancer surgeries, and other types of effective cancer care across different local healthcare markets.
Given these variations – there is a broad consensus that care is clearly suboptimal. In some instances, there may be “overuse” of un-necessary treatments – but for many critical conditions, the problem of “under use” is far more prevalent and problematic. Patients often fail to get evidence-based treatments.
This is why policymakers and private payers turned to P4P: it has tremendous face validity.
The notion is simple: align the incentives for better care. In the old “fee-for-service” world (which is still dominant in cancer care), you got paid more to do more. Providers generally didn’t get paid more to do “better” (that is, to be more evidence based).
It works in most industries: higher quality providers of services get to charge more, deliver better services
While P4P has been around for a very long time, it has gotten substantial traction in healthcare over the past 10 years
So have the efforts with P4P gone? These headlines tell the tale.
Even as of August, 2014, in the NEJM – the latest disappointing story on P4P from England on hospital care – it didn’t work.
If you synthesize the broader field of P4P and its impact: the evidence is underwhelming. It doesn’t appear to work.
See sheet 1 in attached Excel file. Numbers are fudged a bit from Arnie’s paper, but starting and final rates are exact.
Targeted conditions include: AMI, CHF, PNA (excluded hip/knee and COPD)
Data for odds ratios are found in Sheet 8 of the “Data for UCSF talk” excel file.
Sheet 2 in accompanying data file.
From Irini’s unpublished HCAHPS trends paper.
Sheet 1 in accompanying data file.
The conversation on P4P has changed. It is no longer – does P4P work in healthcare? Because the alternatives are not palatable:
Alternative #1 – status quo – not sustainable
Alternative #2 – purely shift risk-taking to providers (i.e. bundled payments, capitation) without substantial quality targets. While this is more viable (and where we are heading) in the short run, lack of P4P and lack of robust quality targets will create substantial political and clinical problems with such models.
Therefore, the real question we need to ask ourselves is: what do we need to do to get P4P to work?