Dr Jonathan B Perlin President, Clinical Services and Chief Medical Officer, HCA (USA) on 'Learning healthcare and clinical leadership in an accountable environment'
1. 2/29/16
1
Learning Healthcare & Clinical Leadership
in an Accountable Environment
Jonathan B. Perlin, MD, PhD, MSHA, MACP, FACMI
President, Clinical Services Group and Chief Medical Officer
HCA / Hospital Corporation of America
Chair, American Hospital Association, 2015
Clinical Professor of Medicine & Biomedical Informatics, Vanderbilt University
Adjunct Professor of Health Administration, Virginia Commonwealth University
Contact: Jonathan.Perlin@HCAHealthcare.com
It’s a Question of Quality Conference
London, England– February 25, 2016
1. US: Accountability, Transparency and Health IT
2. Transformation: Case-Studies from HCA US
– Learning & Improving at-Scale: The REDUCE MRSA Trial
– Clinical Leadership: Billion-Dollar Babies
3. Harvesting the (Big) “Data-Dividend”
Overview
3. 2/29/16
3
• Numerous organizations issue reports ranking and rating hospitals.
• Use CMS and proprietary data
• Issue reports, provide websites, offer recognition
• Some provide a forum for customer/patient reviews
• Some are “pay to play”
• Organizational missions include:
• Consumer advocacy
• Consumer education
• Purchaser decision-making
• Product sales
• Consulting services sales
Third-Party Assessments
L
Third-Party Assessments
Transparency has driven better performance . . . and confusion!
Organizational Responses:
- Learning
- Celebrating Success
4. 2/29/16
4
BREAKING: CMS and health insurers unveil first set of
standardized quality measures
By Melanie Evans | February 16, 2016
The CMS and America's Health Insurance Plans, health plans' trade group,
announced a new agreement to standardize measures of quality for the nation's
doctors. Officials say the measures are necessary as payers and consumers seek
to shop for high-quality care.
The agreement—which outlines seven sets of quality measures to be used
across public and private payers—is the first to be announced by the Core Quality
Measures Collaborative, which includes the CMS, AHIP, the American Academy of
Family Physicians and the National Partnership for Women and Families. The
National Quality Forum, an endorsement body for industry quality standards, was
a technical adviser.
The announcement comes as industry stakeholders and policymakers
struggle with how best to identify and reward high-quality healthcare. Measures of
quality are increasingly tied to how much doctors and hospitals are paid and HHS
said that by 2018 half of Medicare spending outside of managed care would be
under contracts with rewards and potential penalties for quality. . .
Tuesday, February 16, 2016
MU is a Programme in the “HITECH” (Health Information
Technology for Economic & Clinical Health) of ARRA:
• An “Interstate Highway Program” for Health Information
Technology
– Envisioned Interoperable Electronic Health Records
• Offered Incentives to Eligible Hospitals & Providers
(Doctors) to adopt Electronic Health Records over a rolling
four year period to end NLT 2017
– Over $30 BILLION in incentives have been distributed
– BTW, Hospitals not achieving MU would ultimately
experience (incapacitating) payment penalties
Health IT: The “Meaningful Use” (MU) Programme
5. 2/29/16
5
Earning Incentives / Avoiding Penalties Requires
Hospitals & Providers to:
• Use “Certified” Electronic Health Records (EHR)
incorporating Specified Data Standards
• Progressively Increase Use of EHR for
– Physician Order Entry (e.g., labs, medications, imaging)
– Demonstrate interoperability and information exchange
by sharing patient records with other providers and with
patients
– Providing certain public health data to authorities
– Demonstrating use of clinical decision support
– Submitting specified electronic measures of care quality
“Meaningful Use” (MU)
Crossing the Digital Divide . . .
6. 2/29/16
6
Crossing the Digital Divide . . . Physician Practices
The Data Dividend: Learning Healthcare
EHR Use
Decision
Support
Data
Genera9on
Advanced
Analy9cs
>50% of Doctors and
>95% Hospitals met
HITECH Meaningful Use *
* ONC ( hKp://dashboard.healthit.gov/quickstats/quickstats.php )
7. 2/29/16
7
EHR Use
Decision
Support
Data
Genera9on
Advanced
Analy9cs
The Data Dividend: Learning Healthcare
1. US: Accountability, Transparency and Health IT
2. Transformation: Case-Studies from HCA US
– Learning & Improving at-Scale: The REDUCE MRSA Trial
• Making “Meaningful Use” Meaningful
– Clinical Leadership: Billion-Dollar Babies
3. Harvesting the (Big) “Data-Dividend”
Overview
8. 2/29/16
8
Addressing a Significant Problem:
• In the United States, ~ 1 in 20 patients hospitalized will
develop a healthcare-associated infection (HAIs)
~ 1.9 – 2.1 million patients infected
~ 80,000 die
~ $20- 30 billion avoidable healthcare expenditures
• Methicillin-resistant Staphylococcus aureus (MRSA),
and other Staphylococcus aureus, account for
approximately 25% of all deaths from HAIs
The REDUCE MRSA Trial – Background
* Updated from: Klevins RM et al, Pub Hlth Rep, 2007;122:160-6.
What Might be Learned Across a Large Health System ?
InternaZonal
Anchorage
NW GA
Northeast C
Terre Haute
Idaho Falls
Columbus
Atlanta
Panhandle**
Orleans
Idaho Falls
Tallahassee
W
W
W
Las Vegas
W
W
San Jose
W
Western Idaho
Utah
Southern
California
Utah
W
Denver
W
W San Antonio
AusZn
Dallas/FW
Houston
Kansas City
Oklahoma City
W
W
Corpus ChrisZ
Brownsville
W
Wichita
W
W
San Antonio
AusZn
Dallas/FtW
Wichita
El Paso
New
Central
Louisiana
LafayeKe
C
No. VA
C
Richmond C SW VA C
Frankfort C SW VA Frankfort Frankfort
Tampa
North Central Florida
Treasure Coast
E
E
Palm Beach
Dade
Broward
Jacksonville
Columbus
Panhandle
Terre
Haute
Middle GA
Trident/Charleston
Grand Strand Augusta ChaKanooga
SW VA
American Group
NaZonal Group
Central London
Western
Idaho
San Jose
Houston
Kansas City
NW GA
Atlanta
Oklahoma City
Nashville
25 million pa9ent contacts annually
Approximately 5% of major hospital
services in U.S.:
• Admissions > 1.6 million
• PaZent Days > 7.6 million
• Deliveries > 0.25 million
• Total Surgeries > 1.3 million
• ED Visits > 8 million
Ø 170 Hospitals, 120 Freestanding Surgery
Centers, > 850 Physician PracZces, >200 Urgent
Care Centers in 23 states and London
Ø Hospitals range from complex terZary referral
and academic medical centers to urban and
suburban community medical centers
Ø ~ 215,000 employees, including ~
72,000 nurses and 30,000 allied health
professionals
Ø > 50,000 affiliated physicians, including
> 3,300 employed physicians and pracZZoners
Ø More than 38,000 licensed beds
9. 2/29/16
9
Pragmatic Research – Implemented in course of routine care, in routine
setting, not dedicated research unit
Comparative Effectiveness study – Comparing (three) competing “best
practices” to determine what is truly “best”
Cluster-Randomized (by hospital) trial design
1. Screen & Isolate: Screen every patient and implement barrier
isolation, if MRSA positive ( HCA’s “base case” with SIR < 0.7 )
2. Targeted Decolonization: Screen, and if MRSA-positive, isolate
and decolonize (using chlorhexidine antimicrobial soap and
mupurocin nasal ointment)
3. Universal Decolonization: Decolonize all patients on admission
to ICU
Could MU in HCA Provide a Platform for REDUCE MRSA Trial?
Aggregate: 43 Hospitals; 74,356 pa9ents; 282,803 pa9ent-days
Arm 1
16 Hospitals
(23 ICUs)
N = 23,480
Arm 2
14 Hospitals
(22 ICUs)
N = 24,752
Arm 3
13 Hospitals
(29 ICUs)
N = 26,024
16 Hospitals
N = 23,480
13 Hospitals
N = 22,105
13 Hospitals
N = 26,024
1 Hospital (2 ICUs)
withdraws
As
Randomized
As Treated
REDUCE MRSA: Timeline & Enrollment
Jan 2010 Apr 2010 Sep 2011
Baseline
12 month
Phase
In
Interven9on
18 month
HITECH (MU I)
11. 2/29/16
11
Decolonization reduces all blood stream infections (BSIs)
by 44% and MRSA by 37%
• For every 99 patients decolonized, 1 BSI was avoided
• Set a new standard for reducing BSIs in ICUs
• Policy: Demonstrated that (9) state-mandated
screening were expensive and inappropriate
REDUCE MRSA: Study Findings
• Agency for Healthcare Research and
Quality
• CDC Prevention Epicenters Steering
Committee
• Harvard Pilgrim Health Care
Institute / Harvard Medical School
• Hospital Corporation of America
• Rush University
• University of California Irvine
REDUCE MRSA: Discussion
Fostering a Learning Health System
• REDUCE MRSA notable not only for its outcomes, but
for its methods:
• Speed: Did not take one hospital 64 years to amass the
power of the study – it took 43 hospitals 18 months
• Implementation: Not conducted by a single-purpose research
team, but by nurses and infection prevention professionals
during routing patient care
• Setting: Did not occur in a controlled research unit, but within
community hospitals across the country, embedded in routine
care . . .
• Because of interoperable health information, REDUCE MRSA
efficiently answered real-world questions, in real-world
environments, that generalize to real-world situations
12. 2/29/16
12
• Size: 33,000 Patients (vs. 75,000)
• Efficiency: ALLHAT $80M (vs. $3M, including supplies)
• Length of Study: 10 years (vs. 1½ years)
IT-Related Capabilities from Meaningful Use Investment:
1. Standard information platform
• Same EHR system in every hospital (MU Stage 1)
• Conventions to assure semantic interoperability (conceptual
consistency)
2. Aggregation of data into one repository
• n.b., Current work from enterprise clinical data warehouse
3. Normalization of non-standardized data
4. Continuous data quality assessment and feedback
5. Analysis of aggregate data in situ assuring privacy &
security
• i.e., Secure “sandbox” for analytics; no transmission beyond
organization
What Made REDUCE MRSA Possible?
13. 2/29/16
13
0.671 0.686
0.648
0.52
0.388
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Q3 2012 Q4 2012 Q1 2013 Q2 2013 Q3 2013
Universal Decoloniza9on
Phased In
42% Reduction
(3Q12 - 3Q13)
StandardizedInfectionRatio
Source: National Healthcare Safety Network (NHSN)
HCA’s Standardized Infection Ratio for
ICU Central Line-Associated Blood Stream Infection
$170,000/1,000
$19,720,000
Infect Control Hosp Epidmemiology.
2014; 35(S3):S23-S31
The REDUCE MRSA Dividend . . .
Are Learning Opportunities:
- Rare ?
- Subtle ?
- Esoteric ?
Case Study 2: Billion Dollar Babies
14. 2/29/16
14
Another Significant Question:
When is Term Really ‘Term’?
37 to 39 weeks babies
are generally robust,
but is there a scien9fic
basis to say they are
equally robust within
range?
• HCA, along with
March of Dimes (MOD)
and American College
of Obstetricians and
Gynecologists (ACOG)
“called the ques9on”
• Does contemporary
clinical prac9ce =
op9mal outcomes?
• What we did
• Over a 90-day period, 27 hospitals
collected outcomes for deliveries at
> 37 weeks gestaZon
• 17,794 deliveries met criteria
• What we found
• 37 ≠ 38 ≠ 39 weeks
• Why we did it
• ACOG suggested MOD approach
HCA to study quesZon at scale
• HCA commiKed to using data to
change pracZce, if indicated
• HCA delivers ~ 250,000 babies yearly
in 110 hospitals, represenZng ~ 6%
of all U.S. births
*Clark et al, Am J Obstet Gynecol
17.8
8
4.6
0
2
4
6
8
10
12
14
16
18
20
37 Weeks 38 Weeks 39+ Weeks
% NICU Admissions
Weeks of Gesta9on
Elec9ve Term Delivery and
NICU Admission*
Preventing Adverse Neonatal Outcomes
15. 2/29/16
15
• Morbidity
• Lung immaturity, respiratory distress
• Unstable vitals, nutriZonal challenge
• Social disrupZon of birth experience
• Long-Term Morbidity
• Delayed development and weaker
school performance
• Excess Resource U9liza9on
• Unnecessary NICU days and use of
highly skilled nurses and physicians
• More frequent checkups amer
discharge; lost work Zme for parents
17.8
8
4.6
0
2
4
6
8
10
12
14
16
18
20
37 Weeks 38 Weeks 39+ Weeks
% NICU Admissions
Weeks of Gesta9on
Elec9ve Term Delivery and
NICU Admission*
*Clark et al, Am J Obstet Gynecol
Preventing Adverse Neonatal Outcomes
Clinical Behavior Change:
• Trial 2:
• Hard stop (Group 1)
• Peer-review model (Group 2)
• Clinical discussion (Group 3)
• Study demonstrated ‘hard stop’
yielded best results in reducing
early, elec9ve ‘pre-term’ delivery
• MoD, insurers and CMS now
promoZng prevenZng pre-term &
‘hard stop’ as a recommended
pracZce
Transformative Clinical Leadership:
Moving from Knowledge to Practice
*Clark et al, Am J Obstet Gynecol, 2010
November 2010
16. 2/29/16
16
Learning opportunities are ubiquitous
Organizational learning has to be intentional
• EHR’s don’t yet automate “pattern recognition” (i.e., relationship
between gestational age, complications, type of complications and cost
of potentially avoidable services)
Organizational improvement also has to be intentional
• Must obligate to using evidence
Clinical behavior change requires clinical leadership
• Compelling data change discussion from religion to evidence
Learning / Improvement Opportunities
1. US: Accountability, Transparency and Health IT
2. Transformation: Case-Studies from HCA US
– Learning & Improving at-Scale: The REDUCE MRSA Trial
– Clinical Leadership: Billion-Dollar Babies
3. Harvesting the (Big) “Data-Dividend”
Overview
17. 2/29/16
17
What if the results of REDUCE MRSA (of the strategies
compared) were already present in data generated by
previous care?
• What if trials could have been performed “in silica?”
• In 18 minutes, not 18 months ? (REDUCE)
• In 9 minutes, not 90 days ? (Pre-term delivery)
What other answers to pressing questions (cost, quality,
precision medicine, policy) might exist in the “collective
memory” of our healthcare services?
How do we harvest the “Data Dividend?”?
What If REDUCE MRSA Didn’t Require 18 Months?
34
Data sizes:
US Library of Congress printed material: 10 Tb
HCA: 120 Petabytes (12,000 Libraries of Congress)
Google: Best Guess 15 Exabytes
(1.5million Libraries of Congress, 100x HCA)
Planet volumes:
Moon: 2.2E-2
Jupiter: 1.4e3 (63,000 moons)
Sun 1.4E6 (63 million moons)
Units:
Extabyte: 10^18 bytes
Petabyte: 10^15 bytes
Terrabyte: 10^12 bytes
Gigabye: 10^9 bytes
Note that the Library of Congress has at least 3 petabytes of digital content
What Could be Learned from the “Digital Dividend?”
18. 2/29/16
18
Care Informs Care: A Learning Health System
* National Research Council, National Academies of Science, “Toward Precision Medicine:
Building a Network for Biomedical Research and a New Taxonomy of Disease.” 2011; p2.
*
Knowledge
Care Data
Wisdom
Care Changes Care: An Improving Health System
Afferent
Efferent
• Study Design & Question:
• Large, two-arm,cluster randomized, pragmatic comparative-
effectiveness trial in 50 HCA hospitals (~300,000 patients) to
assess the value of chlorhexidine bathing and MRSA
decolonization in adult patients on non-critical care units
• Key Outcomes:
− Unit-associated acquisition of MDROs
− Bloodstream infections: all pathogens
• Additional Outcomes:
− Urinary tract infections: all pathogens
− Contaminated blood cultures
− Infectious readmissions: all pathogens
− Emergence of resistance among key pathogens
− Cost assessment
Active BAthing To Eliminate (ABATE) Infection Trial
> 600,000
19. 2/29/16
19
The “Data Dividend”
. . . a Learning & Improvement Engine
The Full Dividend of Transparency,
Accountability & Clinical Leadership:
A “Learning Health System”
20. 2/29/16
20
APPENDIX
• Major Studies in Progress
• Bibliography
Planned Research Porqolio – Selected Studies
Project HCA Contribu9on Funder Poten9al Impact
1
ABATE: Universal decolonizaZon of
inpaZents not in ICUs
50 affiliated hospitals
~600,000 paZents
NIH
Recommend new standard inpaZent
care to prevent infecZons
2
INSPIRE: Develop and test materials
and methods to improve empiric
anZbioZc therapy
60 HCA hospitals
NIH
Reduce unnecessary and risky use of
broad spectrum anZbioZcs
3
SWAPOUT: Test iodophor nasal
ointment as a component of ICU
decolonizaZon regimen
120 affiliated hospitals
~220,000 ICU paZents
CDC and
Clorox
Improve the standard decolonizaZon
regimen by limiZng risk of resistance
to current agents
4
CLUSTER: Test impact of early
detecZon on size of infecZon
outbreaks in hospitals
80 hospitals CDC
Allow hospitals to mount early
response to outbreaks.
Fully automated system reduces
burden on staff.
5
iMobile: Evaluate effect of secure
messaging among hospital teams
15 hospitals AHRQ
Improve paZent outcomes through
efficient communicaZon
6
Sen9nel System: Assess the safety of
marketed medical products
All HCA hospitals FDA
Improve naZonal capability to
idenZfy risks of medicaZons,
biologics, and blood products
7
PAICAP: Assess ability of CMS data to
monitor the impact of policies to
prevent healthcare associated
infecZons
All HCA hospitals AHRQ
Inform naZonal policy regarding
quality monitoring and payment
incenZve systems