Watson Health 15 Top Health Systems Study, 2018 - 10th edition | April 23, 2018 - Truven Health
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Watson Health 15 Top Health Systems Study, 2018 10th edition | April 23, 2018
IBM Watson Health™ 75 Binney Street Cambridge, MA 02142 800-525-9083 ibm.com/watsonhealth Watson Health 15 Top Health Systems Study, 2018; 10th edition © Copyright IBM Corporation 2018. All rights reserved. IBM, the IBM logo, ibm.com, Watson Health, and 100 Top Hospitals are trademarks of International Business Machines Corp., registered in many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies. Printed and bound in the United States of America. The information contained in this publication is intended to serve as a guide for general comparisons and evaluations, but not as the sole basis upon which any specific conduct is to be recommended or undertaken. The reader bears sole risk and responsibility for any analysis, interpretation, or conclusion based on the information contained in this publication, and IBM shall not be responsible for any errors, misstatements, inaccuracies, or omissions contained herein. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage and retrieval system, without permission in writing from IBM Watson Health. ISBN: 978-1-57372-475-3
Introduction
Contents
Welcome to the 10th edition of the
03 Introduction Watson Health 15 Top Health Systems study.
09 2018 15 Top Health Systems
award winners This 2018 study from IBM Watson Health™
11 Findings marks another milestone in the 100 Top Hospitals®
21 Methodology program’s rich history: a full decade of publishing
37 Appendix A: an annual quantitative study designed to shine
Health system winners and a light on the nation’s highest-performing
their member hospitals health systems.
41 Appendix B:
The top quintile: Highest- Our research of US health system performance
performing health systems began with the same goal that has driven each
43 Appendix C: study since the beginning of the 100 Top Hospitals
Methodology details program: To identify top performers, and also
57 Appendix D: deliver insights that may help healthcare systems
All health systems in study better focus their improvement initiatives on
achieving consistent, balanced, and sustainable
high performance.
Truven Health Analytics® was acquired by Health systems do not apply for our 15 Top Health
IBM in 2016 to help form a new business, Systems selection process, and winners do not pay
Watson Health. to market their honor.
Illuminating achievement for a
value-based world
Our research is based on clinical, operational,
financial, and patient perception-of-care measures
that form a balanced scorecard. For 10 years,
the health systems achieving excellence on our
scorecard inherently set attainable benchmarks for
others in the industry to aspire to over time.
Providing these measures of success may
be especially important today as we see the
healthcare landscape continuing to evolve from
fee-for-service toward value-based care models,
with many health systems paying closer attention
to population health management and system-
wide alignment of performance.
3The Watson Health By finding ways to take balanced performance to
15 Top Health Systems the next level, the winners of our 15 Top Health
scorecard results are divided Systems award are identifying opportunities to
into two separate sections deliver healthcare value to patients, communities,
that graphically illustrate: and payers. The performance levels achieved by
––A health system’s these systems may motivate their peers to use
performance and data, analytics, and benchmarks to close their
improvement versus peer performance gaps.
health systems
––Cross-system performance
alignment of member Delivering a robust and transparent assessment
hospitals
We have designed this study to provide a view
of health system performance across multiple
dimensions: how they stand compared to peers
and high performers (whole-system performance),
where they stand in the evolution of their own
cultures of performance improvement (relative
long-term improvement and rate of improvement),
and the achievement of cross-system performance
alignment (member hospital performance).
These collective insights may be used by health
system leaders to adjust continuous improvement
targets, enrich the collaboration of member
hospitals, and track system-wide alignment toward
common performance goals.
To maintain the 15 Top Health Systems study’s
integrity and avoid bias, we use public data sources
and explain the methodologies we use to calculate
outcome metrics. This approach supports inclusion
of systems across the country and facilitates
consistency of definitions and data.
Our national balanced scorecard, based on Kaplan
and Norton’s concept1, is the foundation of our
research. It is comprised of key measures of
performance: inpatient and extended care quality,
operational efficiency, financial health*, and
patient experience. The composite score derived
from these measures reflects top performance in
hospital-based care, management, and leadership.
* In this study, measures of financial health are reported for information-only purposes because public, audited financial statements are not available for all US health systems.
4 IBM Watson HealthIn addition, to support consideration of the Comparing the performance of our
differences in scale among systems, the study 2018 winners to nonwinners
categorizes the nation’s systems into three groups: Using the measures in our national balanced
large, medium, and small health systems. This scorecard, this year’s 15 Top Health Systems study
produces benchmarks that are comparable and revealed important differences between award
action-driving across similar systems. winners and their nonwinning peers.
Our study’s highest-performing systems:
Yielding a measure of leadership excellence
–– Had lower inpatient mortality and fewer patient
Since 1993, the 100 Top Hospitals program has
complications, considering patient severity
sought to shed light on the efficacy of innovative
leaders. The methodology is aimed at identifying –– Delivered care that resulted in fewer HAIs
leaders who can transform an organization by
–– Had lower 30-day readmission rates*
pinpointing improvement opportunities and
adjusting goals for key performance domains. –– Sent patients home sooner
We believe that higher composite scores on
–– Provided faster emergency care
the balanced scorecard typically indicate more
effective leadership and a consistent delivery –– Kept episode-of-care expenses low, both
of value. in-hospital and through the aftercare process
–– Scored higher on patient ratings of their overall
The leadership of today’s health systems, including
hospital experience
the board, executive team, and medical staff
leadership, is responsible for ensuring all of their
Our study projections also indicate that if the
system hospitals are performing at similarly high
benchmarks of performance established by our
levels in both the short and long term. The 15 Top
2018 winners were achieved by all US health
Health Systems study and analytics provide a view
systems we studied, the following would be true:
of that enterprise performance alignment.
–– Over 60,000 additional lives could be saved
in-hospital
Providing new insight into clinical quality
–– Over 31,000 additional patients could be
For this 2018 15 Top Health Systems study, we complication-free
added a newly available measure of clinical quality:
healthcare-associated infections (HAIs). Since –– 16% fewer infections would be acquired by
there is public interest in tracking and preventing hospital patients
hospital-acquired infections, we used the HAI data ––The typical patient could be released from the
reported by the Centers for Medicare & Medicaid hospital almost a half day sooner and would
Services to analyze hospital performance and have 5.6% fewer expenses related to the
provide national benchmarks in this area. complete episode of care than the median
patient in the US
–– Patients could spend 40 minutes less time in
hospital emergency rooms per visit
* 30-day mortality was slightly higher for winning health systems in this year’s study.
5Understanding the similarities The analysis on the previous page is based on
and differences between applying the difference between study winners
high and low performers can and nonwinners to Medicare patient counts. If the
help provide benchmarks for same standards were applied to all inpatients, the
the industry. The findings we impact would be even greater.
assemble for this study provide
examples of excellence, as For more details about this study’s findings and
evidenced in several additional the achievements of the 15 Top Health Systems,
published studies2 - 24. please see the Findings section of this document.
Welcoming your input
The 100 Top Hospitals program works to ensure
that the measures and methodologies used in our
studies are fair, consistent, and meaningful. We
continually test the validity of our performance
measures and data sources. In addition, as part of
our internal performance improvement process, we
welcome comments about our study from health
system, hospital, and physician executives. To
submit comments, visit 100tophospitals.com.
Showcasing the versatility of the
100 Top Hospitals program
The 15 Top Health Systems research is one of three
major annual studies of the Watson Health 100 Top
Hospitals program. To increase understanding of
trends in specific areas of the healthcare industry,
the program includes:
–– 100 Top Hospitals and Everest Award studies
Research that annually recognizes the
100 top-rated hospitals in the nation based
on a proprietary, balanced scorecard of
overall organizational performance, and
identifies those hospitals that also excel at
long-term rates of improvement in addition
to performance
–– 50 Top Cardiovascular Hospitals study
An annual study introduced in 1999 that
identifies hospitals demonstrating the highest
performance in hospital cardiovascular
services for four important patient groups:
heart attack, heart failure, coronary artery
bypass graft, and percutaneous coronary
intervention
6 IBM Watson Health–– 15 Top Health Systems study
An annual study introduced in 2009 that
The 2018 study
provides an objective measure of health analyzed 338
system performance overall and offers insight
into the ability of a system’s member hospitals
health systems and
to deliver consistent top performance across 2,422 hospitals that
the communities they serve, all based on our
national health system scorecard are members of
health systems.
In addition to the major studies, customized
analyses are also available from the 100 Top
Hospitals program, including custom benchmark
reports. Our reports are designed to help
healthcare executives understand how their
organizational performance compares to peers
within health systems, states, and markets.
100 Top Hospitals program reports offer a
two-dimensional view of both performance
improvement over time, applying the most
current methodologies across all years of data to
identify trends, as well as the most current year
performance.
You can read more about these studies, order
customized reports, and view lists of all winners by
visiting 100tophospitals.com.
About IBM Watson Health
Each day, professionals throughout the health
ecosystem make powerful progress toward a
healthier future. At IBM Watson Health, we help
them remove obstacles, optimize efforts, and
reveal new insights to support the people they
serve. Working across the landscape, from payers
and providers to governments and life sciences,
we bring together deep health expertise; proven
innovation; and the power of artificial intelligence
to enable our customers to uncover, connect, and
act as they work to solve health challenges for
people everywhere.
For more information, visit ibm.com/watsonhealth.
7Note that the order of health systems in the
2018 following tables does not reflect performance
rating. Systems are ordered alphabetically.
15 Top Health For full details on these peer groups and the
process we used to select the winning benchmark
Systems award health systems*, see the Methodology section of
this document.
winners
The Watson Health 100 Top Hospitals® program
is pleased to present the 2018 Watson Health
15 Top Health Systems.
15 Top Health Systems award winners
Large health systems (> $1.85 billion) Location
Mayo Foundation Rochester, MN
Mercy Chesterfield, MO
Sentara Healthcare Norfolk, VA
St. Luke’s Health System Boise, ID
UCHealth Aurora, CO
Medium health systems ($800 million - $1.85 billion) Location
Aspirus Wausau, WI
HealthPartners Bloomington, MN
Mercy Health - Cincinnati Cincinnati, OH
Mission Health Asheville, NC
TriHealth Cincinnati, OH
Small health systems (< $800 million) Location
Asante Medford, OR
CHI St. Joseph Health Bryan, TX
Maury Regional Health Columbia, TN
Roper St. Francis Healthcare Charleston, SC
UPMC Susquehanna Health System Williamsport, PA
* To see a full list of Winners Through the Years, visit truvenhealth.com/Products/100-Top-Hospitals/Program-Info/15-Top-Health-Systems/Winners-Through-the-Years.
9Note: In Tables 1 through 4, data for the
Findings 15 Top Health Systems award winners is labeled
“Benchmark,” and data for all health systems,
The Watson Health 15 Top Health Systems study excluding award winners, is labeled “Peer group.”
profiles the top-performing health systems* in the In columns labeled “Benchmark compared
country. According to publicly available data and with peer group,” we calculated the actual and
our transparent methodologies, these industry percentage difference between the benchmark
leaders appear to be successfully addressing the hospital scores and the peer group scores.
challenge of deploying innovative clinical and
operational approaches to multiple hospital sites 15 Top Health Systems had better survival rates**
to achieve consistent top performance. ––The winners had 14.6% fewer in-hospital
deaths than their nonwinning peers,
For 10 years, the 15 Top Health Systems study considering patient severity (Table 1)
has followed the results achieved by leading –– Mortality results for medium health systems
health systems and published numerous showed the greatest difference between
examples of the benchmark systems’ clinical winners and nonwinners, with 15.9% fewer
and operational excellence. The study is more deaths among benchmark health systems
than a list of accomplishments; it is a tool US (Tables 2 - 4)
health system leaders can use to help guide their
own performance improvement initiatives. By 15 Top Health Systems had fewer
highlighting what the highest-performing leaders patient complications**
around the country are doing well, we create
aspirational benchmarks for the rest of the industry. –– Patients treated at the winning systems’
member hospitals had significantly fewer
complications, with rates 17.3% lower than
How the winning systems compared at nonwinning system hospitals, considering
to their peers patient severity (Table 1)
In this section, we show how the 15 Top Health –– Large health systems had the greatest
Systems performed within their comparison groups difference between winners and nonwinners,
(large, medium, and small systems), compared with 20.9% fewer complications (Tables 2 - 4)
to nonwinning peers. In addition, we identify
some key findings among comparison groups. For
performance measure details and definitions of
each comparison group, see the Methodology
section of this document.
* To be defined as a health system in this study, an organization must have at least two short-term, general, acute care hospitals with separate Medicare provider identification
numbers. Systems with multiple hospital facilities reporting under one provider ID are profiled as a single hospital in the Watson Health 100 Top Hospitals® study.
** Mortality and complications index values cannot be compared among the three different comparison groups because they are normalized by comparison group.
1115 Top Health Systems had fewer healthcare- 15 Top Health Systems had mixed results on
associated infections* longer-term outcomes
A new ranked measure in the 2018 study, Several patient groups are included in the 30-
healthcare-associated infections (HAIs)**, day mortality and readmission extended care
captures information about the quality of inpatient composite metrics. The mean 30-day mortality
care. Based on nationwide data availability, we rate includes heart attack (AMI), heart failure (HF),
built a composite measure of HAI performance pneumonia, chronic obstructive pulmonary disease
at the system level, considering up to six HAIs, (COPD), and stroke patient groups. The mean 30-
depending on assigned comparison group. The day readmission rate includes AMI, HF, pneumonia,
six reported HAIs are: methicillin-resistant total hip arthroplasty and/or total knee arthroplasty
staphylococcus aureus (MRSA-bloodstream), (THA/TKA), COPD, and stroke patient groups.
central line-associated blood stream infections,
catheter-associated urinary tract infections, 30-day mortality results
clostridium difficile (C. diff), surgical site infections
–– In this year’s study, the winning systems had
(SSIs) following colon surgery, and SSIs following
a higher mean 30-day mortality rate than the
an abdominal hysterectomy.
nonwinning systems, due to slightly higher
–– Among all types of systems, winners overall rates among large and small systems
had a more favorable composite median HAI (Tables 1 - 4)
index value than nonwinner peers, at 0.77
–– Small health systems displayed the largest
versus 0.92, respectively; this reflects 16.2%
gap between winners and nonwinning peers
fewer infections occurring at the 15 Top Health
on 30-day mortality (13.4% versus 12.9%),
Systems compared to other peer systems
while medium health system winners had
(Table 1)
performance the same as nonwinning peers,
–– Small health system winners and nonwinners at 12.7%, which was also the lowest median
showed the most dramatic difference on 30-day mortality value observed in this year’s
HAI performance: winners had a median HAI study (Tables 1, 3, and 4)
composite index value of 0.64, which was
26.5% lower than the median HAI index score 30-day readmission results
at nonwinning systems (0.87) (Tables 2 - 4)
–– Winning health systems had lower 30-day
readmission rates than their nonwinning peers
nationally (0.6 percentage points lower) and for
all comparison groups (Table 1)
–– Small winning systems had the best mean
30-day readmission rate (14.4%) among all
comparison groups and outperformed their
nonwinning peers by the greatest margin,
0.9 percentage points (Tables 2 - 4)
* HAI index values cannot be compared among the three different comparison groups because they are normalized by comparison group.
** As developed by a unit of the Centers for Disease Control and Prevention, the National Healthcare Safety Network, and reported by the Centers for Medicare & Medicaid Services
(CMS) in the public Hospital Compare data set.
12 IBM Watson HealthPatients treated in 15 Top Health Systems 15 Top Health Systems hospitals had lower
hospitals returned home sooner* Medicare spending per beneficiary episode costs
–– Winning systems had a median average length –– Overall, winning systems had a 5.6% lower
of stay (ALOS) of 4.4 days, nearly a half day Medicare spending per beneficiary (MSPB)
shorter than the nonwinner median of 4.9 days index than nonwinners (Table 1)
(Table 1)
–– Medium health systems showed the greatest
––The ALOS difference between winners difference between winners and nonwinners
and nonwinners was consistent across all with an 11.8% lower MSPB index (Table 2)
comparison groups, with benchmark systems
–– Medium winning systems also had the lowest
discharging patients 0.4 days sooner
average MSPB index (0.88) among the
(Tables 2 - 4)
comparison groups (Table 2)
Patients spent less time in 15 Top Health Systems
Patients rated 15 Top Health Systems hospitals
emergency departments
higher than peer system hospitals
The mean emergency department (ED) throughput
composite metric measures the amount of time –– Winning systems had a 2.3% higher overall
spent in the ED. It includes median time from ED score on the Hospital Consumer Assessment of
arrival to ED departure for admitted ED patients, Healthcare Providers and Systems (HCAHPS),
and median time from ED arrival to ED departure which tells us that patients treated by
for non-admitted ED patients. members of the top health systems reported a
better overall hospital experience than those
–– Overall, winning systems had, on average, 40
treated in nonwinning peer hospitals (Table 1)
minutes shorter ED wait times per patient than
nonwinners (Table 1) –– Small health system winners had the best
HCAHPS score (270.3) among the comparison
––The greatest difference between winning
groups (Tables 2 - 4)
systems and their peers was in the medium
health systems comparison group, with ––The small winning systems also had the biggest
medium system winners averaging 40.2 lead over nonwinning peers, with an HCAHPS
minutes less time spent in the ED per patient score that was 3.1% higher (Tables 2 - 4)
visit than nonwinners; the range of time
saved was between 34.8 and 40.2 minutes
(Tables 2 - 4)
* ALOS cannot be compared among the three different comparison groups because values are normalized by comparison group.
13Table 1. National health system performance comparisons (all systems)
Performance measure Medians Benchmark compared with peer group
Winning Nonwinning Difference Percent Comments
benchmark peer group difference
health of US Health
systems systems
Inpatient mortality index1 0.88 1.03 -0.15 -14.6% lower mortality
Complications index1 0.85 1.02 -0.18 -17.3% fewer complications
Healthcare-associated infection (HAI) index 2
0.77 0.92 -0.15 -16.2% fewer infections
30-day mortality rate (%) 3
12.9 12.7 0.1 n/a4
higher 30-day mortality
30-day readmission rate (%) 3
14.7 15.3 -0.6 n/a4
fewer 30-day readmissions
Average length of stay (ALOS) (days)1 4.4 4.9 -0.4 -8.8% shorter stays
Emergency department (ED) throughput measure2 180.3 220.3 -40.0 -18.2% less time to service
Medicare spending per beneficiary (MSPB) index 2
0.94 0.99 -0.06 -5.6% lower episode cost
Hospital Consumer Assessment of Healthcare 270.1 264.0 6.1 2.3% better patient experience
Providers and Systems (HCAHPS) score2
1. Mortality, complications and average length of stay based on Present on Admission (POA)-enabled risk models applied to MedPAR 2015 and 2016 data (ALOS 2016 only).
2. HAI, ED measure, MSPB and HCAHPS data from CMS Hospital Compare Jan 1, 2016 - Dec 31, 2016 data set.
3. 30-day rates from CMS Hospital Compare July 1, 2013-June 30, 2016 data set.
4. We do not calculate percent difference for this measure because it is already a percent value.
NNote: Measure values are rounded for reporting, which may cause calculated differences to appear off.
Table 2. Large health system performance comparisons
Performance measure Medians Benchmark compared with peer group
Benchmark Peer group Difference Percent Comments
health of US health difference
systems systems
Inpatient mortality index1 0.89 1.00 -0.11 -11.0% Lower mortality
Complications index 1
0.82 1.04 -0.22 -20.9% Fewer complications
HAI index 2
0.79 0.93 -0.14 -14.9 Fewer infections
30-day mortality rate (%)3 12.8 12.7 0.1 n/a4 Higher 30-day mortality
30-day readmission rate (%)3 14.6 15.3 -0.7 n/a4 Fewer 30-day readmissions
ALOS (days) 1
4.4 4.9 -0.4 -8.9% Shorter stays
ED throughput measure 2
189.9 229.6 -39.7 -17.3% Less time to service
MSPB index 2
0.94 0.99 -0.05 -5.5% Lower episode cost
HCAHPS score2 270.1 264.0 6.0 2.3% Better patient experience
1. Mortality, complications, and ALOS based on POA-enabled risk models applied to MEDPAR 2015 and 2016 data (ALOS 2016 only).
2. HAI, ED measure, MSPB, and HCAHPS data from CMS Hospital Compare Jan. 1, 2016 - Dec. 31, 2016 data set.
3. 30-day rates from CMS Hospital Compare July 1, 2013 - June 30, 2016 data set.
4. We do not calculate percent difference for this measure because it is already a percent value.
Note: Measure values are rounded for reporting, which may cause calculated differences to appear off.
14 IBM Watson HealthTable 3. Medium health system performance comparisons
Performance measure Medians Benchmark compared with peer group
Benchmark Peer group Difference Percent Comments
health of US health difference
systems systems
Inpatient mortality index1 0.85 1.01 -0.16 -15.9% Lower mortality
Complications index 1
0.85 1.03 -0.18 -17.0% Fewer complications
HAI index 2
0.77 0.92 -0.15 -16.5% Fewer infections
30-day mortality rate (%)3 12.7 12.7 0.0 n/a4 No difference in 30-day mortality
30-day readmission rate (%)3 14.7 15.2 -0.5 n/a4 Fewer 30-day readmissions
ALOS (days) 1
4.5 4.9 -0.4 -7.6% Shorter stays
ED throughput measure 2
181.0 221.2 -40.2 -18.2% Less time to service
MSPB index 2
0.88 1.00 -0.12 -11.8% Lower episode cost
HCAHPS score2 269.2 265.0 4.2 1.6% Better patient experience
1. Mortality, complications, and ALOS based on POA-enabled risk models applied to MEDPAR 2015 and 2016 data (ALOS 2016 only).
2. HAI, ED measure, MSPB, and HCAHPS data from CMS Hospital Compare Jan. 1, 2016 - Dec. 31, 2016 data set.
3. 30-day rates from CMS Hospital Compare July 1, 2013 - June 30, 2016 data set.
4. We do not calculate percent difference for this measure because it is already a percent value.
Note: Measure values are rounded for reporting, which may cause calculated differences to appear off.
Table 4. Small health system performance comparisons
Performance measure Medians Benchmark compared with peer group
Benchmark Peer group Difference Percent Comments
health of US health difference
systems systems
Inpatient mortality index1 0.99 1.03 -0.04 -3.4% Lower mortality
Complications index 1
0.83 1.00 -0.17 -17.1% Fewer complications
HAI index 2
0.64 0.87 -0.23 -26.5% Fewer infections
30-day mortality rate (%)3 13.4 12.9 0.5 n/a4 Higher 30-day mortality
30-day readmission rate (%)3 14.4 15.3 -0.9 n/a4 Fewer 30-day readmissions
ALOS (days) 1
4.5 4.9 -0.4 -8.7% Shorter stays
ED throughput measure 2
177.3 212.2 -34.8 -16.4% Less time to service
MSPB index 2
0.95 0.99 -0.04 -4.0% Lower episode cost
HCAHPS score2 270.3 262.2 8.1 3.1% Better patient experience
1. Mortality, complications, and ALOS based on POA-enabled risk models applied to MEDPAR 2015 and 2016 data (ALOS 2016 only).
2. HAI, ED measure, MSPB, and HCAHPS data from CMS Hospital Compare Jan. 1, 2016 - Dec. 31, 2016 data set.
3. 30-day rates from CMS Hospital Compare July 1, 2013 - June 30, 2016 data set.
4. We do not calculate percent difference for this measure because it is already a percent value.
Note: Measure values are rounded for reporting, which may cause calculated differences to appear off.
15Appendix A.
16 IBM Watson Health
Winning health system results
For a list of all hospitals included
In Table 5, we provide the 15 Top
Health Systems values for each of
in each winning health system, see
the study’s performance measures.
Table 5. Winning health system performance measure results
Winning system name Mortality Complications HAI index2 30-day 30-day ALOS1 ED MSPB HCAHPS
index1 index1 mortality readmission measure index4 score4
rate3 rate3 mean
minutes4
Large health Mayo Foundation 0.82 1.23 0.7 11.9 14.6 4.5 166.6 0.92 278.9
systems
Mercy 0.89 0.72 0.8 12.8 14.9 4.4 174.6 0.96 269.9
Sentara Healthcare 0.87 0.82 0.9 12.9 15.2 4.7 235.8 0.94 270.1
St. Luke's Health System 0.96 0.74 0.8 12.5 14.1 4.1 206.3 0.88 268.9
UCHealth 0.91 0.88 0.9 12.9 14.1 4.1 189.9 0.98 271.5
Medium health Aspirus 0.98 0.71 0.6 13.3 14.8 3.9 154.6 0.69 269.2
systems
HealthPartners 0.87 1.08 0.8 12.7 14.7 4.3 174.8 0.88 268.7
Mercy Health - Cincinnati 0.68 0.85 0.5 12.0 15.4 4.6 210.0 1.00 271.6
Mission Health 0.82 1.02 1.0 13.0 13.2 4.5 195.1 0.87 275.2
TriHealth 0.85 0.83 0.8 12.3 14.7 4.5 181.0 0.99 267.3
Small health Asante 0.99 0.53 0.6 13.1 14.3 4.5 185.0 0.91 270.3
systems
CHI St. Joseph Health 0.73 0.88 0.5 14.1 15.3 4.6 166.6 0.98 273.0
Maury Regional Health 1.05 0.51 0.5 13.6 15.5 3.9 177.3 0.95 269.0
Roper St. Francis Healthcare 1.06 0.83 1.1 12.0 13.9 4.8 159.2 0.99 276.9
UPMC Susquehanna Health System 0.74 1.09 0.7 13.4 14.4 4.3 180.3 0.79 267.0
1. Mortality, complications, and ALOS based on POA-enabled risk models applied to MEDPAR 2015 and 2016 data (ALOS 2016 only).
2. HAI, ED measure, MSPB, and HCAHPS data from CMS Hospital Compare Jan. 1, 2016 - Dec. 31, 2016 data set.
3. 30-day rates from CMS Hospital Compare July 1, 2013 - June 30, 2016 data set.
Note: Mortality, complications, HAI and ALOS measures cannot be compared across comparison groups because they are normalized by comparison group.Top and bottom quintile results ––They had somewhat lower mean 30-day
We divided all the health systems in this study into mortality rates (0.4 percentage points lower;
performance quintiles, by comparison group, based includes AMI, HF, pneumonia, COPD, and
on their performance on the study’s measures. stroke patients)
In Table 6, we have highlighted differences ––They had lower mean 30-day readmission
between the highest- and lowest-performing rates (14.8% versus 15.5%; includes AMI,
quintiles by providing their median scores on the HF, pneumonia, THA/TKA, COPD, and
study performance measures. (See Appendix B stroke patients)
for a list of the health systems included in the
top-performance quintile and Appendix D for all ––They had dramatically lower mean ED wait
systems included in the study.) times, with an average difference of 62
minutes per ED patient (24.1% less than the
The top quintile systems outperformed their lowest bottom quintile)
quintile peers in the following ways: ––They were more efficient, releasing patients
––They had much better patient outcomes: almost one day (0.9) sooner than the lowest
19.7% lower mortality and 5.7% lower performers and at a 5.3% lower MSPB index
complications ––They scored 11.3 points higher on the HCAHPS
––They had fewer occurrences of HAIs in their overall patient rating of care
facilities: a 25.8% smaller median HAI index
value (0.8) in the top performance quintile
Table 6. Comparison of health systems in the top and bottom quintiles of performance1
Performance measure Top quintile Bottom quintile Difference Percent Top versus bottom
median median difference quintile
Inpatient mortality index2 0.89 1.11 -0.22 -19.7% Lower mortality
Complications index2 0.99 1.05 -0.06 -5.7% Fewer complications
HAI index 3
0.8 1.0 -0.3 -25.8% Fewer infections
30-day mortality rate (%) 4
12.6 12.9 -0.4 n/a 5
Lower 30-day mortality
30-day readmission rate (%) 4
14.8 15.5 -0.8 n/a 5
Fewer 30-day readmissions
ALOS 2
4.5 5.4 -0.9 -16.1% Shorter stays
ED measure mean minutes3 195.0 256.9 -62.0 -24.1% Less time to service
MSPB index3 0.96 1.01 -0.05 -5.3% Lower episode cost
HCAHPS score 3
269.0 257.7 11.3 4.4% Better patient experience
1. Top and bottom performance quintiles were determined by comparison group and aggregated to calculate medians.
2. Mortality, complications, and ALOS based on POA-enabled risk models applied to MEDPAR 2015 and 2016 data (ALOS 2016 only).
3. HAI, ED measure, MSPB, and HCAHPS data from CMS Hospital Compare Jan. 1, 2016 - Dec. 31, 2016 data set
4. 30-day rates from CMS Hospital Compare July 1, 2013 - June 30, 2016, data set.
5. We do not calculate percent difference for this measure because it is already a percent value.
Note: Measure values are rounded for reporting, which may cause calculated differences to appear off.
17Test metrics: Reported for information only Excess days in acute care measures
Every year, we evaluate the 15 Top Health Systems The newest set of measures available from CMS
study and explore whether new measures would in the Hospital Compare data set are the excess
enhance the value of the analysis we provide. days in acute care (EDAC) measures for AMI and
For this 2018 study, we are testing several HF. CMS defines “excess days” as the difference
new performance measures that update basic between a hospital’s average days in acute
standards of inpatient care and expand the care and expected days, based on an average
balanced scorecard across the continuum of care. hospital nationally. Days in acute care include
These metrics were not used in the ranking and days spent in an ED, a hospital observation unit,
selection of winning health systems. or a hospital inpatient unit for 30 days following
a hospitalization. The data period in our study for
If you would like to provide feedback on these measures is the same as for the other 30-day
the following proposed measures, email metrics for specific patient conditions: three years,
100tophospitals@us.ibm.com. combined (July 1, 2013 - June 30, 2016).
30-day all-cause, hospital-wide 90-day episode-of-care payment measure
readmission measure Another measure recently made available in the
We are continuing to publish the hospital-wide, Hospital Compare data set is the 90-day episode-
30-day readmission measure, which CMS is of-care payment metric for primary, elective
publicly reporting in the Hospital Compare data set THA/TKA. Like the other 30-day episode-of-
to provide an overall readmission comparison, for care payment measures, CMS calculates risk-
information only. However, we rank on a composite standardized payments associated with a 90-day
score based on the publicly available individual episode of care, compared to an average hospital
patient groups. The data period for the hospital- nationally. The measure summarizes payments
wide readmission measure is July 1, 2015 - for patients across multiple care settings, services,
June 30, 2016. and supplies during the 90-day period, which
starts on the day of admission. The data period for
30-day episode-of-care payment measures this measure combines three years, April 1, 2013 -
We are continuing to publish risk-standardized March 31, 2016.
payments associated with 30-day episode-of-care
measures for three patient groups that are now 90-day complication measure
being published by CMS in the Hospital Compare Along with the THA/TKA 90-day payment
data set. These measures capture differences in measure recently made available in Hospital
services and supplies provided to patients who Compare data, CMS is also publishing a THA/
have been diagnosed with AMI, HF, or pneumonia. TKA 90-day complication measure. This measure
According to the CMS definition of these measures, calculates a risk-standardized complication rate for
they are the sum of payments made for care and THA/TKA procedures using the occurrence of one
supplies starting the day the patient enters the or more of the subsequent complications within
hospital and for the next 30 days. In our study, the the specified timeframes. The data period for this
data period for these measures is the same as measure combines three years, April 1, 2013 -
for the other 30-day metrics for specific patient March 31, 2016 (complications are listed on the
conditions: three years, combined (July 1, 2013 - following page).
June 30, 2016).
18 IBM Watson Health–– AMI, pneumonia, or sepsis/septicemia/shock –– On all reported CMS episode measures of cost,
during or within seven days of index admission winning systems consistently outperformed
their nonwinner peers; the difference was
–– Surgical site bleeding, pulmonary embolism,
greatest for THA/TKA 90-day episode payment
or death during or within 30 days of index
(5.8%)
admission
–– Winning health systems performed better
–– Mechanical complication or periprosthetic joint
on the AMI and HF 30-day EDAC measures,
infection/wound infection during or within 90
showing that patients spent fewer days than
days of index admission
expected in the ED, in observation, or back in
the hospital after an initial index acute care
See the CMS website for measure methodology25.
stay (5.2 days under the expected amount
for AMI EDAC and 8.5 under expected for HF
Table 7 shows the national performance of
EDAC); whereas nonwinning peers averaged
benchmark and peer health systems on the
7.6 and 8.1 days more than expected (EDAC
test metrics.
values are reported as excess days per 100
discharges)
This year, the 15 Top Health Systems winners
outperformed nonwinning peers on all test –– On the 90-day THA/TKA complications rate,
measures, which is an interesting finding given that another CMS measure of outcomes extended
these are independent variables not used in the outside the hospital stay, winning systems had
selection of the winners. 0.3 percentage points lower complication rate
than peers
–– Benchmark systems had stronger performance
on 30-day, hospital-wide readmissions (14.8%
versus 15.4% for peers)
Table 7. Information-only measures – Health system performance comparisons (all classes)
Performance measure Medians Benchmark compared with peer group
Benchmark Peer group Difference Percent Comments
health systems of US health difference
systems
30-day, hospital-wide readmission rate1 14.8 15.4 -0.63 n/a3 Fewer 30-day readmissions
AMI 30-day episode payment 1
$22,798 $23,204 -$406 -1.8% Lower episode cost
HF 30-day episode payment1 $16,023 $16,376 -$353 -2.2% Lower episode cost
Pneumonia 30-day episode payment 1
$17,117 $17,318 -$201 -1.2% Lower episode cost
AMI 30-day excess days in acute care 1
-5.2 7.6 -12.81 n/a 3
Fewer excess days
HF 30-day excess days in acute care 1
-8.5 8.1 -16.62 n/a 3
Fewer excess days
THA/TKA* 90-day episode payment 2
$20,653 $21,931 -$1,278 -5.8% Lower episode cost
THA/TKA* 90-day complications rate2 2.4 2.7 -0.30 n/a3 Fewer complications
1. 30-day measures from CMS Hospital Compare July 1, 2013 - June 30, 2016 data set.
2. 90-day measures from CMS Hospital Compare April 1, 2013 - March 31, 2016 data set.
3. We do not calculate percent difference for these measures because it can be a negative number or is already a percent value.
* Primary, elective total hip arthroplasty and total knee arthroplasty.
19Financial metrics –– Overall, benchmark health system
We continue to publish the financial measures performance is better than nonwinning peers,
each year for information only, as audited financial both on operating margin (1.2 percentage
statements are not available for all systems points higher among winners) and long-term
included in the study*. These measures are not debt-to-capitalization ratio (LTD/cap)
included in the ranking and selection of benchmark (0.1 lower ratio among winners)
health systems. –– Notably, medium health system winners had
a much higher operating profit margin than
Results for included systems are found in nonwinners (8.7% versus 2.4%) and showed
Table 8 below. a greater difference on the LTD/cap (0.2
versus 0.4)
Table 8. Information-only – Financial performance
Performance measure Health system Medians Difference
comparison group
Benchmark health systems Peer group of US Benchmark compared
health systems with peer group
Operating margin All systems 4.2 3.0 1.2
(percentage)
Large 4.3 3.6 0.7
Medium 8.7 2.4 6.3
Small 4.0 2.2 1.8
Long-term debt-to- All systems 0.3 0.3 -0.1
capitalization ratio
(LTD/cap) Large 0.3 0.3 0.0
Medium 0.2 0.4 -0.1
Small 0.3 0.3 -0.1
Note: Data sourced from audited 2016 financial reports via dacbond.com, emma.msrb.org, yahoo.brand.edgar-online.com, and sec.gov.
* A total of 84.2% of parent and independent systems published audited financial statements for 2016. Subsystems that are members of a larger “parent” health system generally
do not have separate audited financial statements. This translated into 65.1% of all in-study health systems with available financial reports.
20 IBM Watson Health–– Ranking systems on each of the performance
Methodology measures by comparison group
–– Determining the 15 top performers (five in
Watson Health 15 Top Health Systems is a each comparison group) from the health
quantitative study that annually identifies 15 systems’ overall rankings, based on their
US health systems with the highest overall aggregate performance (sum of individual
achievement on a balanced scorecard. weighted measure ranks)
The health system scorecard is based on the The following section is intended to be an overview
100 Top Hospitals® national balanced scorecard of these steps. To request more detailed information
methodologies and focuses on four performance on any of the study methodologies outlined here,
domains: inpatient outcomes, extended outcomes, email us at 100tophospitals@us.ibm.com or call
operational efficiency, and patient experience. 800-525-9083.
This 2018 health systems study includes nine
measures that provide an objective comparison Building the database of hospitals
of health system performance using publicly Like all the 100 Top Hospitals studies, the 15 Top
available data. The health systems with the highest Health Systems study uses only publicly available
achievement are those with the highest ranking on data. The data for this study primarily came from:
a composite score based on these nine measures.
–– Medicare Provider Analysis and Review
To analyze health system performance, we include (MEDPAR) data set*
data for short-term, acute care, nonfederal US –– Centers for Medicare & Medicaid Services
hospitals, as well as cardiac, orthopedic, women’s, (CMS) Hospital Compare data set
and critical access hospitals (CAHs) that are
members of the health systems. We use MEDPAR patient-level demographic,
diagnosis, and procedure information to calculate
The main steps we take in selecting the top 15 mortality, complications, and length of stay (LOS)
health systems are: by aggregating member hospital data to the health
–– Building the database of health systems, system level. The MEDPAR data set contains
including special selection and exclusion criteria information on the approximately 15 million
Medicare patients discharged annually from US
–– Identifying which hospitals are members of acute care hospitals. In this year’s study, we used
health systems the most recent two federal fiscal years (FFYs) of
–– Aggregating the patient-level and hospital- MEDPAR data available (2015 and 2016), which
level data from member hospitals and included Medicare Advantage health maintenance
calculating a set of performance measures at organization encounters. The 100 Top Hospitals
the system level program has used the MEDPAR database for
many years. We believe it to be an accurate and
–– Classifying health systems into comparison reliable source for the types of high-level analyses
groups based on total operating expense performed in this study.
* The MEDPAR data years quoted in 100 Top Hospitals research are based on an FFY, a year that begins on October 1 of each calendar year and ends on September 30 of the
following calendar year. FFYs are identified by the year in which they end (for example, FFY 2016 begins October 1, 2015, and ends September 30, 2016). Data for all CMS Hospital
Compare measures is provided in calendar years, except the 30-day rates. CMS publishes the 30-day rates as three-year combined data values. We label these data points based
on the end date of each data set. For example, July 1, 2013 - June 30, 2016, is named “2016.”
21We used the CMS Hospital Compare data The recalibrated models were used in producing
set published in the third quarter of 2017 for the risk-adjusted inpatient mortality and
healthcare-associated infection (HAI) measures, complications indexes, based on two years of
30-day mortality rates, 30-day readmission rates, MEDPAR data (2015 and 2016). The severity-
Medicare spending per beneficiary (MSPB) index, adjusted LOS was produced based on MEDPAR
and Hospital Consumer Assessment of Healthcare 2016 data.
Providers and Systems (HCAHPS) patient
perception-of-care data26. Present-on-admission coding adjustments
From 2010 through 2016, we have observed a
We also used the 2016 Medicare cost reports, significant rise in the number of principal diagnosis
published in the federal Hospital Cost Report and secondary diagnosis codes that do not have
Information System (HCRIS) third-quarter 2017 a valid POA indicator code in the MEDPAR data
data set, to create our proprietary database for files. Since 2011, an invalid code of “0” has been
determining system membership based on “home appearing. This phenomenon has led to an artificial
office” or “related organization” relationships rise in the number of complications that appear to
reported by hospitals. The cost reports were also be occurring during the hospital stay. See Appendix
used to aggregate member hospital total operating C for details.
expense to the system level. This data was
used to classify health systems into three To correct for this bias, we adjust MEDPAR record
comparison groups. processing through our inpatient mortality and
complications risk models and LOS severity-
Risk- and severity-adjustment models adjustment model as follows:
The IBM Watson Health™ proprietary risk- and
1. Original, valid (Y, N, U, W, or 1) POA codes
severity-adjustment models for inpatient mortality,
assigned to diagnoses were retained
complications, and LOS have been recalibrated for
this study release using FFY 2015 data available 2. Where a POA code of “0” appeared, we took
in the all-payer Watson Health Projected Inpatient the next four steps
Database (PIDB). The PIDB is one of the largest
a. We treated all diagnosis codes on the
US inpatient, all-payer databases of its kind,
CMS exempt list as “exempt,” regardless
containing approximately 23 million inpatient
of POA coding
discharges annually, obtained from approximately
5,000 hospitals, which comprise more than 65% b. We treated all principal diagnoses as
of the nonfederal US market. Watson Health risk- “present on admission”
and severity-adjustment models take advantage
c. We treated secondary diagnoses where the
of available present-on-admission (POA) coding
POA code “Y” or “W” appeared more than
that is reported in all-payer data. Only patient
50% of the time in the Watson Health all-
conditions that are present on admission are
payer database as “present on admission”
used to determine the probability of death,
complications, or the expected LOS. d. All others were treated as “not present”
22 IBM Watson HealthHospital exclusions Health system exclusions
After building the database, we exclude hospitals Health systems are excluded if:
that would have skewed the study results.
–– One or more required measures are missing*
Excluded from the study were:
–– Fewer than 50% of member hospitals have valid
–– Certain specialty hospitals (children’s,
POA coding
psychiatric, substance abuse, rehabilitation,
cancer, and long-term acute care) –– Fewer than 50% of member hospitals have valid
data for any one or more required measures**
–– Federally owned hospitals
–– Hospitals not located within the 50 states After all system exclusions were applied, 338
(such as those in Puerto Rico, Guam, and the individual health systems were included in the
US Virgin Islands) 2018 study.
–– Hospitals with Medicare average LOS longer
NOTE: CMS does not publish MSPB measures for
than 30 days in FFY 2016
Maryland hospitals due to a separate payment
–– Hospitals with no reported Medicare patient agreement. For this reason, we substituted the
deaths in FFY 2016 comparison group median, and winner-excluded
Maryland health systems that had no reported
–– Hospitals that had fewer than 60% of patient
MSPB measure to allow Maryland health systems
records with valid POA codes
to remain in the study. If a Maryland health system
included hospitals in other states, we winner-
Cardiac, orthopedic, women’s hospitals, and CAHs
excluded them when more than 50% of their
are included in the study, if they are not excluded
member hospitals had no reported MSPB measure.
for any other criteria listed above.
In addition, specific patient records are
also excluded:
–– Patients who were discharged to another
short-term facility (this is done to avoid
double-counting)
–– Patients who were not at least 65 years old
–– Rehabilitation, psychiatric, and substance
abuse patients
–– Patients with stays shorter than one day
After all exclusions were applied, 2,422 individual
hospitals were included in the 2018 study.
* For composite measures (HAI, 30-day mortality, 30-day readmissions), the exclusion is applied ONLY if all individual measure comprising the composite are missing.
- For HAI, different numbers of individual measures were required depending on the comparison group (five for large and medium systems; three for small systems). A system not
meeting the minimum was excluded. See Appendix C for details.
- In systems where one or more individual 30-day mortality or 30-day readmission rates were missing, BUT NOT ALL, we calculated a median value for each, by comparison
group, and substituted the median for the missing value.
** This rule was not applied to the HAI composite, which followed different exclusion logic. See Appendix C for details.
23Identifying health systems To analyze health system performance, we
To be included in the study, a health system must aggregate data from all of a system’s included
have at least two short-term, general, acute hospitals. In the methodology summary tables in
care hospitals with separate Medicare provider this section, we provide specific details about the
identification numbers. The minimum of two calculations used for each performance measure
hospitals must be met after hospital exclusions and how these measures are aggregated to
have been applied. In addition, we also include determine system performance.
any cardiac, orthopedic, women’s hospitals, and
CAHs that passed the hospital exclusion rules After all exclusions were applied and parent
cited on the previous page. For the 2018 study, systems identified, the final 2018 study group
we identified the “parent” system by finding the included 338 health systems with the profiles
“home office” or “related organization”, as reported outlined in Table 9.
on the hospitals’ 2016 (or 2015) Medicare
cost report.
We identify health systems that have subsystems
with their own reported home offices or related
organization relationships. Both the parent system
and any identified subsystems are treated as
“health systems” for purposes of this study and are
independently profiled. Hospitals that belong to a
parent health system and a subsystem are included
in both for analysis of system performance.
Table 9. 2018 health systems study group
System category Systems Member hospitals Medicare patient Average hospitals Average discharges
discharges, FFY per system per system
2015
Winning systems 15 121 330,760 8.1 22,051
Nonwinning systems 323 2,798 9,368,880 8.7 29,006
Total systems 338 2,919 9,699,640 8.6 28,697
Note: A hospital can be a member of both a parent system and a subsystem of that parent. They will be included in both parent and subsystem member hospital counts.
The total unduplicated hospital count in this study was 2,422 hospitals.
24 IBM Watson HealthClassifying health systems into The nine measures included in the 2018 study, by
comparison groups performance domain, are:
Health system comparison groups
We refine the analysis of health systems by dividing Inpatient outcomes
them into three comparison groups based on 1. Risk-adjusted inpatient mortality index
total operating expense of the identified member
hospitals. This is done to develop more action- 2. Risk-adjusted complications index
driving benchmarks for like systems. For the 2018 3. Mean HAI index
study, the three comparison groups we used are
listed in Table 10. Extended outcomes
4. Mean 30-day risk-adjusted mortality rate
Table 10. Health system comparison groups, defined
(includes acute myocardial infarction [AMI],
Health system Total operating Number of Number of
comparison expense systems in winners
heart failure [HF], pneumonia, chronic
group study obstructive pulmonary disease [COPD],
Large > $1.85 billion 113 5 and stroke)
Medium $800 million - 116 5 5. Mean 30-day risk-adjusted readmission
$1.85 billion
rate (includes AMI, HF, pneumonia, total hip
Small < $800 million 109 5
arthroplasty and/or total knee arthroplasty
Total systems 338 15 [THA/TKA], COPD, and stroke)
Scoring health systems on weighted Operational efficiency
performance measures 6. Severity-adjusted average LOS
Evolution of performance measures
7. Mean emergency department (ED) throughput
We use a balanced scorecard approach, based on
(wait time minutes)
public data, to select the measures we believe to
be most useful for boards, CEOs, and other leaders 8. MSPB index
in the current health system operating environment.
In addition, we continually review trends in the Patient experience
healthcare market, to identify the need for, and
9. HCAHPS score (overall hospital performance)
availability of, new performance measurement
approaches. We welcome feedback from hospital
The data sources for these measures are listed in
and system executives on the usefulness of our
Table 11.
measures and our approach.
As the healthcare industry has changed, our
methods have evolved. Our current measures
are centered on four main components of system
performance: inpatient outcomes, extended
outcomes, operational efficiency, and patient
experience. Measures of financial performance are
also included for information only, as not all health
systems have publicly reported, audited financial
statements.
25Table 11. Summary of measure data sources and data periods
Performance measure Current performance Five-year trend performance
(15 Top Health Systems award selection)
Risk-adjusted inpatient mortality index MEDPAR federal fiscal year (FFY) 2015 and 2016* MEDPAR FFY 2011 - 2016*
Risk-adjusted complications index MEDPAR FFY 2015 and 2016* MEDPAR FFY 2011 - 2016*
Mean HAI index CMS Hospital Compare Calendar Year (CY) 2016 Trend not available
Mean 30-day mortality rate CMS Hospital Compare July 1, 2013 - June 30, 2016 CMS Hospital Compare: Three-year data sets
ending June 30 in 2013, 2014, 2015, 2016
Mean 30-day readmission rate (AMI, HF, CMS Hospital Compare July 1, 2013 - June 30, 2016 CMS Hospital Compare: Three-year data sets
pneumonia, THA/TKA**, COPD, stroke) ending June 30 in 2013, 2014, 2015, 2016
Severity-adjusted ALOS MEDPAR FFY 2016 MEDPAR FFY 2012 - 2016
Mean ED throughput measure CMS Hospital Compare CY 2016 CMS Hospital Compare 2012 - 2016
MSPB index CMS Hospital Compare CY 2016 CMS Hospital Compare 2012 - 2016
HCAHPS score (overall hospital rating) CMS Hospital Compare CY 2016 CMS Hospital Compare 2012 - 2016
* Two years of data are combined for each study year data point.
** Primary, elective total hip arthroplasty and total knee arthroplasty.
Following is the rationale for the selection of our Extended outcomes
balanced scorecard domains and the measures The extended outcomes measures (30-day
used for each. mortality rates for AMI, HF, pneumonia, COPD, and
stroke patients; and 30-day readmission rates for
Inpatient outcomes AMI, HF, pneumonia, THA/TKA, COPD, and stroke
Our measures of inpatient outcomes include patients) help us understand how the system’s
three measures: risk-adjusted mortality index, patients are faring over a longer period27. These
risk-adjusted complications index, and mean HAI measures are part of the CMS Hospital Value-
index. These measures show us how the system Based Purchasing Program and are widely reported
is performing on what we consider to be the most on in the industry. Hospitals with lower values
basic and essential care standards (survival, error- appear to be providing, or coordinating the care
free care, and infection prevention) while treating continuum with better medium-term results for
patients in their hospitals. these conditions.
As systems become more interested in contracting
for population health management, we believe that
understanding outcomes beyond the walls of the
acute care setting is imperative. We are committed
to adding new metrics that assess performance
along the continuum of care as they become
publicly available.
26 IBM Watson HealthYou can also read
