View
2
Download
0
Category
Preview:
Citation preview
W W W. H E A LT H L E A D E R S M E D I A . C O M / I N T E L L I G E N C E
C uncilHEALTHLEADERS MEDIA
Access. Insight. Analysis.
Powered by
PREMIUM REPORTFEBRUARY 2016
THE ANALYTICS CHALLENGE: Gaining Critical Insight into Risk-Based Models
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 2TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
FOREWORD
We are in the epicenter of a large collision. In the midst of EHR
implementations becoming stable, organizations are overwhelmed
with the explosion of new technology and innovation, all while dealing
with decreasing reimbursement for care, increasing government
mandates, and an ever-increasing severity of illness of patients in our
facilities due to higher out-of-pocket expenses. While this reality may
frighten many, the answer to most of these problems lies in the use of
advanced analytics tools.
It is my great honor to introduce HealthLeaders Media’s 2016
Analytics in Healthcare Survey findings. The survey this year included
350 respondents who are senior leaders, as well as operations, clinical,
financial, and information leaders. Survey questions included topics
such as the basis for finance and patient data analytics, applications
of large/complex data sets, need for analytics software due to risk
contracts, uses of clinical and financial analytics now and in the future,
financial and clinical data analytics capabilities, and the top analytics
challenges over the next three years.
A clear finding is the three-year plan of how organizations expect
to use clinical analytics: 30% currently leverage this technology to
populate registries, while 43% expect to do so in three years. In
addition, the use of clinical analytics to develop risk stratification is
moving from 41% today to 65% in three years.
Patient data acquisition is one of the new technology frontiers that
is rapidly evolving. Only 17% of respondents currently use data from
patient health monitors as part of their analytics efforts, though 43%
anticipate doing so within three years.
Despite these findings, we have a long way to go. Our ability to move
from descriptive to predictive analytics is still in its infancy. But this
development will take our healthcare systems to the next level. The
great opportunity to leverage financial and clinical informatics, risk
modeling, and disparate data sources are the solutions organizations
are approaching as we all deal with all these mounting issues.
While our struggles of balancing technology benefits with implementa-
tion challenges will continue in an aggressive fashion over the next few
years, the vision of leveraging advanced analytics in healthcare gives
hope and promise for stable healthcare systems—systems that can
deliver the outcomes every patient desires and deserves.
As we become more proactive in our data analytics, our ability to
forecast disease states, outcomes, and therapies positively changes
the way we care for patients. For at the end of the day, whether we
are involved in direct patient care or in strategic planning and support,
all we do is about providing excellent, high-quality patient care in this
country and making it easier to navigate the healthcare enterprise
every day. And avoiding the large collisions.
Advanced Analytics: The Solution to Many Challenges
Stephen Beck, MD, FACP, FHIMSSChief Medical Informatics Officer
Mercy Health Cincinnati, Ohio
Lead Advisor for this Intelligence Report
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 3TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
Table of Contents
Foreword 2
Principal Conclusions and Recommendations 4
Analysis 6
Case Studies 15
Analytics Project Supports ACO Goals of Reduced Costs, Improved Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Analytics Initiative Identifies Gaps in Care, Reduces Hypoglycemic Event Frequency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Data Collaborative Facilitates Integration of Payer and Clinical Data . . . . . .20
Survey Results 23
This document contains privileged, copyrighted information. If you have not purchased it or are not otherwise entitled to it by agreement with HealthLeaders Media, any use, disclosure,
forwarding, copying, or other communication of the contents is prohibited without permission.
Fig. 1: Use of Financial Analytics Now. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Fig. 2: Use of Financial Analytics Within Three Years . . . . . . . . . . . . . . . 24
Fig. 3: Types of Finance Data Drawn On for Analytics Activity Now . . 25
Fig. 4: Types of Finance Data Drawn On for Analytics Within Three Years . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Fig. 5: Use of Clinical Analytics Now . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Fig. 6: Use of Clinical Analytics Within Three Years . . . . . . . . . . . . . . . . . 28
Meeting Guide 39
Methodology 40
Respondent Profile 41
Fig. 7: Types of Patient Data Drawn On for Analytics Now . . . . . . . . . . 29
Fig. 8: Types of Patient Data Drawn On for Analytics Within Three Years . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Fig. 9: Current Applications for Working With Large/Complex Data Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Fig. 10: Presence of Downside Risk Contracts Prompting Need for Analytics Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Fig. 11: Financial Data Analytics Capabilities . . . . . . . . . . . . . . . . . . . . . . . 33
Fig. 12: Clinical Data Analytics Capabilities. . . . . . . . . . . . . . . . . . . . . . . . . . 34
Fig. 13: Top Data-Related Analytics Challenges Over Next Three Years . 35
Fig. 14: Top Tactical Analytics Challenges Over Next Three Years. . . . . . 36
Fig. 15: C-Suite Title Responsible for Financial Analytics . . . . . . . . . . . . . . 37
Fig. 16: C-Suite Title Responsible for Clinical Analytics . . . . . . . . . . . . . . . 38
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 4TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
DATA POINTS
> The top uses for financial analytics now are determining cost of care (68%), cost containment efforts (66%), and financial risk assessment (61%). Responses for population risk assessment (35%) place it last. (Figure 1)
> For financial analytics use within three years, responses for population risk assessment (60%) show the greatest growth, increasing 25 percentage points, followed by financial risk assessment (up 15 percentage points to 76%). (Figure 2)
> Sixty-four percent of respondents say that the presence of contracts with downside risk has prompted the need for or increased their dependence on analytics software or services, with 16% saying that dependence has led them to acquire analytics software or services, and 23% saying they are investigating doing this. (Figure 10)
> Comparing the use of clinical analytics now with clinical analytics use within three years, the biggest response increases are for assessing population health needs (up 27 points to 73%), lowering cost of care (up 23 points to 72%), and developing risk stratification (up 24 points to 65%). (Figures 5 and 6)
The transition to value-based care and the assumption of greater downside
risk has providers looking to analytical tools to better help them understand
the increasingly complex industry they serve. Among the many data
challenges they face are integrating financial and clinical data, improving EHR
interoperability, and analyzing payer- and patient-related data from a diverse
range of internal and external sources.
However, perhaps one of the biggest challenges they confront is balancing
day-to-day operational needs with long-term strategies that are often
resource-intensive and highly changeable. It will require a deft leadership
touch to appropriately serve today’s short-term demands with industry
transformation looming in the not-too-distant future.
IT investment supports multiple strategies. Investing in the IT function
covers a lot of ground for most healthcare organizations—it keeps
information infrastructure up and running so that day-to-day financial
operations proceed consistently and, increasingly, it also supports a growing
list of activities around tracking and managing clinical performance
and outcomes. These near-term demands often consume considerable
HealthLeaders Media Research Editor-Analyst Jonathan Bees draws on the data, insights, and analysis from this report:
PRINCIPAL CONCLUSIONS AND RECOMMENDATIONS
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 5TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
management energy and funding, but they are a necessary part of an
organization’s strategy for maintaining and updating IT infrastructure.
However, in spite of these pressing near-term demands, new investments
must also be made with a long-term perspective, and it is especially critical
that the organization’s IT function is aligned with its strategic direction as
it relates to taking on risk and delivering value-based care. Of particular
importance is analytics, which affects both the nature of the activity that IT
must support and the need for the multiple sources of data that will be the
basis for analysis. Establishing a solid foundation of analytics knowledge,
skills, and technologies will require persistence and a constant revisiting of
strategic priorities—after all, an organization’s strategic plans probably will
be revised more frequently because care models and payment mechanisms
are currently in flux.
Descriptive, predictive, and prescriptive. There are three distinct types
of analytics capability—descriptive (what has happened), predictive
(what will happen, given past data), and prescriptive (predictive plus
proactive solutions). While the majority of respondent organizations
support descriptive analytics, this retrospective view of data, while useful,
will be insufficient to meet the challenges of taking on greater risk and
delivering value-based care in the future. Healthcare leaders need to invest
in predictive and prescriptive analytics capabilities so that they can make
decisions proactively and take preventive action, in both their financial and
clinical operations. Perhaps most challenging of all, the financial and clinical
groups’ analytical needs must be fulfilled in as close to real time as possible.
Balancing short- and long-term priorities. Although IT planning and
investment should take into account both the organization’s short- and
long-term strategy, near-term requirements around industry consolidation,
ambulatory/outpatient expansion, and value-based care initiatives can
create exceptional demands for resources. Organizations must navigate this
delicate balancing act of addressing short- and long-term priorities so that
they are not forced to choose one at the expense of the other. For analytics,
take a close look at evaluating in-house development versus using outside
services—in-house development may consume more human resources than
is sustainable and, ultimately, outside services might offer a faster and more
cost-effective path to bring applications online.
PRINCIPAL CONCLUSIONS AND RECOMMENDATIONS (continued)
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 6TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
The evolution of analytics from mainly functioning as a finance and
administration tool to an expanded role that includes integrating financial and
clinical data continues to gain momentum. Its use and increasing sophistication,
driven by the steady advance of value-based care and the assumption of
greater downside risk by providers, are trends that will likely accelerate in the
coming years.
According to the 2016 HealthLeaders Media Analytics in Healthcare Survey,
for example, the top uses for financial analytics now (Figure 1)—determining
cost of care (68%), cost containment efforts (66%), financial risk assessment
(61%)—are traditional financial analytics uses. Responses for population
risk assessment (35%) place it at the bottom of the list. However, when
respondents are asked about financial analytics use within three years (Figure
2), responses for population risk assessment (60%) show the greatest growth,
increasing 25 percentage points, although it remains at the bottom of the list.
As the industry moves from fee-for-service to value-based care, providers will
likely find themselves consumed with analyzing payer- and patient-related data
from diverse internal and external sources. The need for analytics tools to make
sense of the disparate sources for both financial and clinical data will only grow.
ANALYSIS
The Rise of Analytics and the New Risk-Based ParadigmJONATHAN BEES
Here are selected comments from leaders regarding their strategy for
improving the effectiveness of their organization’s analytics staff, and whether
they will focus on developing in-house talent or look to hire from outside.
“We are looking for better engagement with clinicians; the analytics staff
aren’t close enough to patient care to know what to look for and how to explain
variation. The focus will be on developing in-house talent.”
—Chief financial officer at a medium health system
“We will use both in-house and outside talent. We are creating a culture of
integrating analytics with a strategic mindset of leveraging and identifying
new opportunities for change.”
—CEO at a small physician organization
“We are moving our independent data analytics teams to imbed them
within finance and clinical quality to improve the contextual analysis of the
information.”
—Chief financial officer at a medium health system
“For financial analytics, we will probably hire from outside. For clinical
analytics, we will focus on internal development. The strategy at the moment
involves waiting for the EHR company to develop an effective data mining
product so that we can obtain information for clinical analytics, and then
training someone in how to use these.”
—Chief compliance officer at a small hospital
WHAT HEALTHCARE LEADERS ARE SAYING
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 7TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
“How do we leverage all of that data into a decision-support infrastructure, or
into a customer relationship management type of infrastructure that allows
us to proactively manage the health of that patient in a way that eventually
will deliver the highest outcomes at the lowest cost? And how do we time
the evolution to these models? Because when you think about it, today the
majority of reimbursement is still from fee-for-service,” says Indranil “Neal”
Ganguly, FCHIME, FHIMSS, CHCIO, vice president and chief information
officer at JFK Health Systems, an Edison, New Jersey–based nonprofit
healthcare system.
“It is in our best interest right now from a financial standpoint to have the
visit, to see the patient. But as we transition over to a world of either pay-
for-performance or risk-based contracting, we’re really looking to minimize
the number of physical visits or interactions. And at that point we need to
rely much more heavily on technology to give us predictive intervention
capabilities to say, ‘Hey, I see a trend here; let me intervene now when it’s
much cheaper to intervene than when the patient is more acute and will
present at a care setting.’ ”
Financial analytics. Along with the 25-point jump in population risk
assessment analytics efforts within three years, financial risk assessment
increases 15 percentage points, moving from 61% to 76% (Figures 1 and 2).
Clearly, providers are anticipating that they will need to apply analytics tools
to manage the increased risk associated with value-based care.
As providers undertake contracts
with increasing levels of downside
risk, their need for advanced
analytics to manage decision-
making and monitor results will
also grow. Sixty-four percent of
respondents say that the presence
of contracts with downside risk
has prompted the need for or
increased their dependence on
analytics software or services
(Figure 10), with 16% saying that
dependence has led them to acquire
analytics software or services and
23% saying they are investigating
analytics software and services.
Looking at the types of finance-related data organizations draw on now for
analytics activity (Figure 3), respondents indicate that Medicare/Medicaid
patient claims data (78%), commercial payer patient claims data (67%),
and internal provider productivity data (59%) are the top data sources. The
prominence of the response for payer claims data suggests that providers are
focusing on the relationship between patient care and revenue.
“I think we’re going to have to get much more predictive in using analytics. Historically, we’ve been using analytics retrospectively, so it’s looking at the past and then trying to make some kind of judgment as to how we should manage the future.”
—Indranil “Neal” Ganguly
ANALYSIS (continued)
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 8TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
ANALYSIS (continued)
The ongoing move to value-based care also influences provider thinking when
it comes to the types of finance-related data organizations draw on within
three years for analytics activity (Figure 4). While the top data types remain
the same as for data drawn on now—Medicare/Medicaid patient claims data
(up 5 points to 83%), commercial payer patient claims data (up 9 points to
76%), and internal provider productivity data (up 7 points to 66%)—it is
noteworthy that the largest increases are found on other data types.
Care partners’ cost data (up 24 points to 41%), payer cost data (up 18
points to 56%), and care partners’ provider productivity data (up 17 points
to 36%) experience the greatest increases in response. The results for the
two care partner data types indicate the importance of the role that the
care continuum plays in delivering value-based care, and the degree to which
providers expect to use analytics to evaluate that data.
Clinical analytics. As one might expect, improving clinical quality (85%) is
the top response by a wide margin for use of clinical analytics now (Figure 5).
Identifying gaps in care (65%) and identifying variations in care (56%) place
a step below in usage. And echoing the results for use of financial analytics
now (Figure 1), responses for assessing population health needs (46%) place
it in the bottom half of responses.
Within three years (Figure 6), respondents indicate that improving clinical
quality remains at the top (up 4
points to 89%), while identifying
gaps in care retains second place
(up 11 points to 76%). Tied for
third are identifying variations in
care (up 17 points to 73%) and
assessing population health needs
(up 27 points to 73%). Other notable
increases in clinical analytics
uses are seen for lowering cost of
care (up 23 points to 72%) and
developing risk stratification (up
24 points to 65%). The majority
of respondents expect to use
clinical analytics for a wide range
of applications—responses are grouped in a relatively tight cluster, with all
but one (populating registries) receiving a response of 65% or higher. These
are the kinds of activities that relate to value-based care and the ongoing
transformation efforts within the industry.
In a similar vein, the leading responses for patient-related data drawn on
now for analytics activity (Figure 7) are clinical data from EHR (78%) and
patient demographics (72%) in the top tier, followed by a second tier right
“You don’t want to automate it so much that people don’t think anymore, and now they say, ‘Oh, you know, I’m not worried about that number because if it was high enough, it would trigger my MEWS score and I would see the alert.’ ”
—Stephen Beck, MD
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 9TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
ANALYSIS (continued)
around 50%: aggregated EHR and patient claims data (51%), patient lab and
imaging data (48%), and patient pharmaceutical data (48%).
Looking ahead three years, the leading data types (Figure 8) are the same,
although the response levels are higher, and the gap between top-tier and
second-tier responses shrinks: clinical data from EHR (up 10 points to
88%), patient demographics (up 8 points to 80%) at the top, followed by
aggregated EHR and patient claims data (up 20 points to 71%), patient
pharmaceutical data (up 20 points to 68%), and patient lab and imaging data
(up 18 points to 66%).
The greatest increase in response is in the third tier, where data drawn from
patient health monitors, such as remote telemetry, increases 26 points to
43%. The double-digit growth for most categories is a reflection of provider
anticipation of population health management requirements.
Descriptive, predictive, and prescriptive analytics. For the purposes
of this survey, analytics capabilities are segmented into three categories:
descriptive (what has happened), predictive (what will happen, given past
data), and prescriptive (predictive plus proactive solutions). Respondents
were asked to evaluate their financial and clinical analytics capabilities by
these categories.
For financial analytics (Figure 11),
84% of respondents indicate that
they have descriptive analytics
capability, while 49% report some
predictive capability, and just 25%
have prescriptive capability. Not
surprisingly, a large majority are
able to analyze activities that have
happened in the past. However,
these results indicate that a substantial portion do not even have the ability
to apply analytics to what has happened, calling into question their ability
to make informed decisions for their organization. Of course, only 4% of
organizations with net patient revenue of $1 billion or more lack such basic
descriptive analytics capability for financial data.
While results for the highest level of analytics represents the smallest
segment, it is encouraging that 25% of respondents have prescriptive
analytics, which will be critical in enabling providers to take timely preventive
action in the future through a degree of software intelligence.
Results for clinical analytics follow a similar pattern (Figure 12), with 83%
reporting descriptive analytics, 35% with predictive analytics, and 26%
“I think absolutely the future is in predictive and prescriptive analytics.”
—Sue Schade, MBA
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 10TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
ANALYSIS (continued)
prescriptive. The response for predictive clinical analytics is lower than that
for predictive financial analytics by 14 points, which is perhaps an indication
of the difficulty of applying analytics to patient health.
Consistent with the results for financial analytics, while a large percentage
of respondents (83%) have support for descriptive clinical analytics, this
indicates that a substantial portion do not have the ability to apply clinical
analytics to what has happened, a somewhat troubling finding. Again, this
advanced capability is greater among billion-dollar organizations.
The use of analytics to understand past activity should be the baseline, and
predictive and prescriptive analytics are advancements necessary to move
healthcare forward.
“I think we’re going to have to get much more predictive in using analytics,”
says Ganguly. “Historically, we’ve been using analytics retrospectively, so it’s
looking at the past and then trying to make some kind of judgment as to how
we should manage the future.
“We have to get down to looking at the data at two levels,” he says. “One is
being able to understand the broad trends in our populations, but then take
that understanding and create logic that we can apply at the individual level.
To say, for example, if we see a growing trend in A1c values increasing, we
know we’ve got an increasing problem of diabetes in our community. Now,
how do we take that information
back down to the individual level and
target collection of data in a way
that will allow us to predict who’s at
risk for becoming a type 2 diabetic?
And how do we stop or mitigate
those risks as far in advance as
possible?
“I believe there are indexes that
people are leveraging, such as LACE,
and there are a number of clinical
predictive algorithms that are out
there,” Ganguly says. The LACE
Index identifies patients who are at
risk for readmission or death within
30 days of discharge; the initials represent four factors: length of stay, acuity
of the admission, comorbidities using the Charlson comorbidity index, and
emergency room visits in the past six months. “Where we get challenged is
[that] they’re relatively new. They’re not super-well-tested, so some people
question the reliability of the predictive models and algorithms. But I think
that the evidence base is growing and, as that grows, these models will
become more and more valid.
“They’re not super-well-tested, so some people question the reliability of the predictive models and algorithms. But I think that the evidence base is growing and, as that grows, these models will become more and more valid.”
—Indranil “Neal” Ganguly
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 11TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
“Predictive analytics is not a new concept,” Ganguly says. “Clinicians have
been doing this for a long time, such as when they stage pressure ulcers
or other things, it’s sort of a common clinical pathway. But now they’re
becoming much more advanced in their utilization. They’re bringing together
more complex tools. But they’re applying them in the bigger chronic areas—
such as heart failure patients, diabetics, and pulmonary patients—where
you’ve got a high cost of treatment. And if you are able to intervene early
enough, you can significantly reduce the cost of care.”
According to Stephen Beck, MD, FACP, FHIMSS, chief medical informatics
officer at Mercy Health, a nonprofit, Catholic health system serving the
Kentucky and Ohio region, predictive analytics doesn’t have to be a complex
undertaking.
“We do this on the inpatient side to look at trends and the numbers, using a
predictive modeling where you know the MEWS [or Modified Early Warning
System] score. You know that based on certain data points you can generate
a score, and this means that patient is at risk for an event—for example, their
heart stopping for a code blue. We performed this on paper, actually, before
we started doing it electronically. And what we found was you could suggest
a new therapy based on one of those measures that would prevent a code
blue,” he says.
“We’re doing more and more of this,”
Beck says. “I believe that the biggest
challenge still comes back to, How
do you address those types of tools
with workflow? Because we know
when you can automate it to the
point where you don’t have to think
about it, then you tend to have good
outcomes. The problem with that
is, you don’t ever want to take the
human brain out of the equation.
“You don’t want to automate it
so much that people don’t think anymore, and now they say, ‘Oh, you know,
I’m not worried about that number because if it was high enough, it would
trigger my MEWS score and I would see the alert,’ ” he says.
“I think absolutely the future is in predictive and prescriptive analytics,” says
Sue Schade, MBA, FCHIME, FHIMSS, LCHIME, chief information officer
at the University of Michigan Hospitals and Health Centers, which is part
of University of Michigan Health System, a nonprofit academic healthcare
system that serves the Michigan and Northern Ohio area. “For example, we’ve
done some predictive analytics around hemodynamic instability. And I would
ANALYSIS (continued)
“As you’re moving toward value-based medicine, or risk-based contracts, you have to figure out how to take better care of your patients. And that’s a very complicated task.”
—Stephen Beck, MD
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 12TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
say that’s definitely in the realm of predictive and prescriptive analytics in
terms of evaluating patients that are decompensating, and determining what
we can know in advance and being able to address it.”
Analytics challenges. The top three data-related challenges respondents
expect to face over the next three years (Figure 13) are integrating internal
clinical and financial data (54%), establishing/improving EHR interoperability
(47%), and integrating external clinical and financial data (41%). These data-
related challenges involve data that originates from multiple sources, which
adds a degree of complexity for providers.
Beck says, “When it comes to using analytics to manage populations, you
may have an empaneled patient that seeks care elsewhere, even though I’m
the PCP. How do I get that information? It’s fine if it’s part of my network on
my EHR, but if it’s outside, how do we get that information back in a timely
fashion and in a way that feeds back into the decision support that I have?
“As an example,” he says, “we’ve worked with Walgreens in our area to ensure
that if they’re giving an immunization, that message is coming back to us so
that we know and can have a record updated. There’s no second-guessing if
we have to do an outreach to the patient regarding a pneumococcal vaccine.
“But I believe the bigger issue is the disparate data systems, the disparate
electronic records in our
communities,” Beck says. “We’re
doing everything that we can to
try to integrate the pieces into
an enterprise data warehouse to
help pull those information pieces
together, yet make the decision
support that comes back to a doctor
much smarter. And it may be as
simple as taking CPT category
II code data or other billing data
directly from the payer. The payer
states that the patient had that
pneumonia vaccine done because
they paid for it.”
Ganguly says he sees interoperability as a lesser problem compared with the
others on the list. “Interoperability is, to some degree, more of an annoyance
than a disaster point. We can exchange data pretty well with most other
systems. I know many of my colleagues are all participating in a variety of
different types of health information exchange initiatives. I think what we’re
not seeing, though, is how we’re getting value out of the data exchange. So
ANALYSIS (continued)
“I will tell you most CIOs kind of cringe at the term big data because it’s one more term that’s taken hold within the industry that everybody grabs on to but probably doesn’t really understand.”
—Sue Schade, MBA
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 13TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
exchanging the data, or aggregating it some way, is less of a problem than
figuring out how to get value out of it and consume some kind of output from
the aggregated data.”
Schade explains that certainly one of the big challenges organizations face is
a multitude of priorities. “Organizations have been very focused on the core
work of EHR, operationally, replacing and installing new systems, and that has
consumed a tremendous amount of IT time. Also, we focus on meaningful use,
and we had to get ready for ICD-10. And during this period I have seen from
an operational perspective just a real increase in the demand around having
good data and being able to do reporting. And I think a lot of organizations
are not necessarily planning ahead for this.
“At the same time,” she says, “there’s a growing demand for deeper analytics
and the emergence of big data—which, I will tell you most CIOs kind of
cringe at the term big data because it’s one more term that’s taken hold
within the industry that everybody grabs on to but probably doesn’t really
understand. And so CIOs like myself are working to make sure that we
are very effective in that core reporting and then—looking more broadly,
especially in an academic medical center—as to what our overall analytics
platform needs to be when you take into account, for us, the tripartite mission
of clinical care, research, and education.”
Tactical challenges. Selecting
the top three tactical analytics
challenges they expect to face
in the next three years (Figure
14), respondents cite a top tier of
overcoming insufficient skills in
analytics (45%) and the need to
deliver timely analysis (45%). The
second-tier responses cited by about
one-third are insufficient funding
in light of other priorities (35%),
picking the right platform for data
and analytics (33%), and insufficient
staff (32%).
Some of the challenges relate indirectly or directly to financial resources,
while others are more closely tied to levels of expertise. Note that the need to
deliver timely analysis is a universal challenge for all industries, although it is
perhaps more acute in healthcare.
Schade says that larger organizations often face challenges related to the
breadth of their organizational data footprint. “I think that you probably have
a lot of organizations similar to ours that have various siloes of data, various
ANALYSIS (continued)
“Exchanging the data, or aggregating it some way, is less of a problem than figuring out how to get value out of it and consume some kind of output from the aggregated data.”
—Indranil “Neal” Ganguly
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 14TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
data warehouses that different parts of the organizations have developed,
with the result that you end up with a more federated type of architecture.
Certainly some organizations have been ahead of the curve and have been
able to develop a more centralized platform.”
Ganguly cites the sheer volume of data available for evaluation. “The amount
of information available is tremendous, and it’s growing exponentially.
The [number of] savvy members of the end-user community, whether it’s
clinicians or financial types, is growing, and they understand that there’s
more they can do with the information. So they’re looking for more reporting,
they’re looking for more tools. Our resources are traditionally limited, so how
do we prioritize? You can’t meet all of the demand. So that’s one challenge.
And then I think the other challenge, and part of this is a demand control
strategy, is how do we begin to educate our data customers to handle some
of the basic analytics themselves?”
The human element. While it may be tempting to view analytics as a
panacea for the many challenges of value-based care, there are some aspects
of healthcare transformation that are resistant to analytical approaches.
Consider, for instance, the human element.
Beck summarizes it this way. “As you’re moving toward value-based medicine,
or risk-based contracts, you have to figure out how to take better care of your
patients. And that’s a very complicated task. The reason it’s so complicated
is you’ve got a complicated population—some respond well to verbal
communication and will do everything the doctor asks them to do; some will
take their medication on time and others will not.
“Using analytics, how can you really determine the difference between a
compliance issue—so I’ve ordered a test and the patient just hasn’t had
it done—versus not ordering the test at all because I missed it or a staff
member missed it? It requires taking the combination of decision-support
and active-management tools that are at the point of care, and outreach
tools that are population health–based. And not just implementing them, but
implementing them in a way to make sure that we’re getting the outcomes
that we’re really looking for.”
Jonathan Bees is research editor-analyst for HealthLeaders Media.
He may be contacted at jbees@healthleadersmedia.com.
ANALYSIS (continued)
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 15TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
CASE STUDY 1
Analytics Project Supports ACO Goals of Reduced Costs, Improved Outcomes
To support its population health
strategy, JFK Health began an
accountable care organization called
JFK Population Health, LLC, as part
of the Medicare Shared Savings
Program.
Before the ACO went live in January
2014, JFK Health’s IT department
began, in October 2013, preparing
to provide analytics support for
the initiative. Since that time, the
department has focused mainly on
building an analytics interface and
establishing a data repository.
“The IT department viewed it as
a good chance to get exposed to
accountable care and some of the
concepts around care coordination,
and how we can support the hospital
and physicians to make sure that we are identifying gaps in care, gaps in
preventive care, and driving to reduce things like unnecessary readmissions,”
says Indranil “Neal” Ganguly, FCHIME, FHIMSS, CHCIO, vice president and
chief information officer at JFK Health.
ACO integration. JFK Population Health has a two-pronged management
structure for facilitating the integration of the ACO into the parent
organization, based on a Finance Committee and a Quality and Operations
Committee. At one point, there was also an IT Committee.
“We used to have an IT Committee, but we disbanded it because the
physicians didn’t really get into the technology piece of the process. So we
just do that on the back end through the hospital,” says Ganguly. “The Finance
Committee requires data because ACOs have a gainsharing component
to them, and they’re looking at the model by which any gains would be
distributed back to the physicians who are meeting their care objectives. The
Quality and Operations Committee is looking primarily at how well we are
optimizing the outcomes for the patients.”
Early challenges. The first year of the analytics project was mainly
concerned with building the infrastructure for data collection and
JFK HEALTH JFK Health is an Edison, New Jersey–based nonprofit healthcare system that serves the central area of the state. It owns and operates JFK Medical Center, a 498-bed acute care hospital, as well as the JFK Hartwyck Nursing and Rehabilitation Centers and the JFK Johnson Rehabilitation Institute, which feature inpatient and outpatient rehabilitation centers and nursing facilities. The organization has approximately 950 medical staff. In 2014, the system reported $514 million in net patient revenue.
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 16TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
CASE STUDY 1 (continued)
aggregation. This aspect offered a number of challenges due to the disparate
systems used by the various members of the ACO.
“We began with a focus on data collection, and this year is the first we’re
actually producing measurements for the providers to look at in terms of
improvement,” says Ganguly. “Collecting the data is quite a challenge because
of the nature of the physician community and the number of systems in use
there. Some physicians are capturing the data electronically to interface to
their systems. Others are capturing it via some level of manual abstraction,
either at the practice or using hospital resources. Ultimately, it’s all being fed
into a repository where we’re running the analytics and then looking at how
the individual practices are doing against the 33 ACO quality measures from
CMS.”
EMRs and data flow. Inevitably, building the data collection infrastructure
was primarily focused on the ACO physicians’ EMRs. Besides facing the usual
EMR compatibility issues, the IT department also found that not all EMRs
were being used to the fullest extent of their capabilities.
“Even though most of the physicians had EMRs, they weren’t all set up to
capture all of the data. Oftentimes they were just set up in a default manner
and they were capturing enough basic data for the physician to bill properly,
but not necessarily capturing all of the data they would need to manage the
care of the patient.”
Multiple data sources. The data collection initiative is not limited to the ACO
alone. JFK Health is building an enterprise data repository that encompasses
ACO provider data, inpatient data from JFK Medical Center, and data from an
approximately 30-hospital health information exchange in which JFK Health
participates.
“We don’t want to be narrowly restricted to what we know about a patient
from their interaction in our environment alone. Because patients are fluid,
they may come to JFK for a certain thing but then they sprain their ankle in
front of a neighboring hospital, they may go there. With an ACO, the primary
care physician especially needs to know what’s going on with that patient.
But if it’s outside of the JFK ecosystem, how do we report that data?”
Physician workflow. There is also a human element to establishing the
analytics platform. While the ACO’s physicians are organized around a
standard set of data collection practices, they also have to adapt to a new
workflow.
“Part of the challenge is just adoption,” says Ganguly. “Obviously, when you’re
first establishing something like an ACO, it’s not the physicians’ entire panel
that’s involved. So they’re delivering care along a certain path for their entire
patient base, and now they have to take a new path for some subset of their
patients because they are enrolled in an ACO.
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 17TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
CASE STUDY 1 (continued)
“How do you build that into their workflow in a way that doesn’t impede
productivity? Some have staff in the office who will look at the data and prep
the physician in advance; other physicians are more hands-on, looking at it
themselves. But [for] the ones that are able to look at it and really understand
it, then the only question is, are they closing the care gaps effectively?”
Work in progress. While basic reporting on the 33 CMS quality measures
is now active—physicians are able to see their performance compared
against their peers within the ACO as well as against national averages—the
analytics initiative is still in the early stages of development, says Ganguly.
“It’s actually one of the challenges we see across the industry when you get
to analytics. If you don’t have the data to begin with, how do you really build a
true analytics environment? And right now, we have so much data, it’s the old
conundrum—we have tons of data but not enough information. And so the
key is we need to collect usable targeted data at the front end that feeds the
necessary analytics to drive quality and cost control, which is where we really
want to go.”
—Jonathan Bees
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 18TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
At Mercy Health, low blood sugar
event rates were running higher
than organizational expectations,
presenting a potential threat to
patient health and outcomes.
“In the fall of 2013, we identified that
there was an issue relative to a large
number of hypoglycemic events,” says
Stephen Beck, MD, FACP, FHIMSS,
chief medical informatics officer at
Mercy Health. “And so we created a
multidisciplinary team that included
nursing, pharmacy, clinical content
or nursing informatics, as well as
physician representation to examine
where our care gaps might be that
were creating the issue.”
Mercy Health captures and tracks reportable events using a third-party
system. Beck says, “The analysis process initially started with our automated
reporting system for reportable events, and out of that data we then used
analytics to determine that there were a higher than expected number of low
blood sugar events.
“There is also a self-reporting aspect to this, so that any time a staff member
notes or considers a risk to the patient, or a potential harm for the patient,
they can denote it very quickly through this reporting package. And then it
gets assigned as a task, because one of the ways that my team uses this is
when a clinical person denotes a risk, we then try to identify a cause.”
Multiple events per patient. Once Mercy Health determined that
hypoglycemic events were running higher than expected, the organization’s
own analytics group, CarePATH Clinical Solutions, began a deep dive on the
data. One of the first revelations was that the high number of reportable
events was being driven not only by the number of patients with single events,
but also by the number of patients having multiple events.
“That’s a multiple event diminishment,” says Beck. “If a patient had a single
MERCY HEALTHMercy Health is a nonprofit, Catholic health system that has 23 hospitals, eight senior housing facilities, and seven home health agencies serving the Kentucky and Ohio region. Formerly known as Catholic Health Partners, the organization has approximately 450 locations providing care and has over 32,000 employees, including over 1,300 employed physicians. Annual net patient service revenue was $3.8 billion in 2014.
CASE STUDY 2
Analytics Initiative Identifies Gaps in Care, Reduces Hypoglycemic Event Frequency
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 19TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
CASE STUDY 2 (continued)
event, that was categorized, but we also looked to see which of those patients
had double events, if they had two low blood sugar readings. And then we
looked to see which ones had three, and which had four, and which had five
and six, and so forth, all the way up to 10. And what we saw is that we were
having significant numbers of four and five events per patient.”
There were two types of hypoglycemic events studied in the initiative. Critical
low blood sugar events, which are defined as having a blood glucose level of
below 50 mg/dL, and below normal events, which are a blood glucose level
of between 50 and 70 mg/dL. Blood glucose information is tracked in the
patient’s EHR.
Identifying the cause. After studying both the critical and below normal
blood sugar level events reported in the EHR, the analytics team came to an
interesting conclusion. The main culprits for the problem were the workflow
for ordering insulin and a disconnect in the process for addressing low blood
sugar monitoring and medication.
Beck describes the workflow issue: “Clinicians were ordering insulin, and not
placing orders for the ‘what if.’ And in this case, the ‘what if ’ was, if there’s
low blood sugar, what do you do? At that time, we had an order panel that
basically said if you order insulin you should order this panel as well, which
says check the blood sugar every so often, and if it goes below this level, then
administer this medication, stop the insulin, and so forth. We’ve always had
these, but they weren’t linked together. And so that was a simple observation
that we were then able to identify and fix, and in February of 2014 we added
this panel and embedded it inside our order set where you order insulin.
“We also identified an additional gap that when a provider ordered insulin as
a single line item and not as part of an order set, there was the possibility that
we were going to miss some additional patients. So we developed an advisory
a few months later, in June of 2014, so that if this treatment panel wasn’t
ordered along with the insulin you would get an alert, basically automating
the ability to order that panel.”
The results. Adoption of the new protocols produced almost immediate
results. Comparing first quarter with second quarter 2014, total critical low
blood sugar events declined 12.0% (916 versus 806 events) and the number
of patients experiencing more than three critical events declined by 12.6%
(190 versus 166 patients). Likewise, total below normal low blood sugar
events declined 7.3% (3,322 versus 3,081 events) and the number of patients
experiencing more than three below normal events declined by 26.1% (1,109
versus 819 patients).
By project end in fourth quarter 2014, total low blood sugar events had been
reduced by 70% compared with the first quarter results.
—Jonathan Bees
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 20TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
CASE STUDY 3
The Michigan Data Collaborative is a
nonprofit organization that maintains
a statewide database of multi-payer
claims and provider EHR data.
Established in 2008, the organization
is based at the University of Michigan
Health System and falls under the
umbrella of the health system’s IT
department. Data management
and analytical services for MDC are
provided by IT department employees.
MDC was created in partnership with
Blue Cross Blue Shield of Michigan as
part of an effort to foster population
health IT infrastructure and improve
care delivery in the state of Michigan.
MDC’s main function is to provide data
and analytics support for healthcare
transformation initiatives across the state, primarily working in
conjunction with the Michigan Primary Care Transformation (MiPCT)
demonstration project, a Center for Medicare & Medicaid Innovation
project that is part of its Multi-Payer Advance Primary Care program.
According to Sue Schade, MBA, FCHIME, FHIMSS, LCHIME, chief
information officer at University of Michigan Hospitals and Health
Centers, “The Michigan Data Collaborative, or MDC, is part of a
relationship we have with Blue Cross Blue Shield of Michigan that goes
back to 2008. Originally, it was established to create a consolidated
claims database for payers in Michigan, and over time we’ve done
additional work on it to include clinical data. It’s a really important
foundation for improving healthcare in Michigan.”
Funding for MDC is project based, currently coming from the MiPCT
project (which itself is funded in part by the CMS Center for Medicare
& Medicaid Innovation), participating payers such as Blue Cross Blue
Shield of Michigan, and the state of Michigan.
UNIVERSITY OF MICHIGAN HEALTH SYSTEM
The University of Michigan Health
System is a nonprofit academic
healthcare system that serves the
Michigan and Northern Ohio area.
It has three acute care hospitals,
40 outpatient locations, 140
clinics, and more than 26,000
faculty, staff, students, trainees,
and volunteers. Its six specialty
centers provide care for cancer,
cardiovascular, depression, eye,
diabetes, and geriatric patients.
In 2014, the system reported
$2.5 billion in operating revenue.
Data Collaborative Facilitates Integration of Payer and Clinical Data
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 21TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
CASE STUDY 3 (continued)
Organization partners and mission. MDC unites payers and providers in
sharing claims and EHR data, which is then aggregated and made available
to physician organizations participating in the MiPCT initiative. There are
currently five payers in the program: Blue Cross Blue Shield of Michigan,
Blue Care Network of Michigan, Medicare, Medicaid, and Priority Health, a
nonprofit health plan. On the provider side, there are currently 37 physician
organizations. The repository contains data on approximately 4 million
covered lives, and has information on roughly 1.3 billion medical claims
processed over the last five years.
Schade says that the MDC currently has the following core objctives:
• Create a full picture of care—regardless of payer
• Measure population-wide quality and outcomes
• Identify high-risk populations with chronic and comorbid conditions
• Help identify interventions and best practices that work
• Help our partners identify and track cost reduction opportunities
“When Blue Cross initially approached us—I wasn’t here at the time—this
partnership was about establishing a consolidated claims database for all
payers,” says Schade. “At the beginning, it focused on claims data for five
payers in Michigan. However, ultimately, they wanted to combine clinical
data with claims data to show whether patients’ health was actually
improving versus only seeing if a patient had a recommended test or visited
a physician. As the payers have evolved, they are now looking at value and
outcomes more. But I think even what we were doing seven years ago was
fairly leading edge.”
Dashboards and reports. MDC provides a broad range of dashboards
and reports based on the integration of payer claims and clinical data. In
addition, Truven Health Analytics, which also participates in the initiative,
contributes data on risk scores, standard costs, and admission information.
It had initially assisted in aggregating the data, which MDS now handles
in-house.
The MDC repository tracks chronic conditions such as asthma, attention
deficit hyperactivity disorder, chronic kidney disease, chronic obstructive
pulmonary disease, coronary artery disease, diabetes, chronic heart failure,
hypertension, and obesity. It records utilization rates for the emergency
department, as well as admissions and readmissions activity in general. It
monitors quality measures for adult chronic care, adult preventive care, and
pediatric care. And it tracks 15 electronic clinical quality measures.
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 22TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
CASE STUDY 3 (continued)
15 electronic clinical quality measures
Diabetes - HbA1c poor control
Diabetes - HbA1c control
Diabetes - LDL-C control
Diabetes - blood pressure control
Asthma action plan
Hypertension - blood pressure control
CVD - blood pressure control
CVD - LDL-C control
BMI
Tobacco use
Colorectal cancer screening
Asthma action plan – pediatrics
BMI – pediatrics
Tobacco use – pediatrics
Depression screening for patients with chronic conditions
The primary data and analytics report users are the MiPCT physician
organizations. However, all of the payers have access to the aggregated
data, as does the University of Michigan Health System.
Looking forward, MDC is expected to add two additional multipayer
initiatives in 2016.
—Jonathan Bees
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 23TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
FIGURE 1 | Use of Financial Analytics Now
Click on these icons to dig deeper.
What does your organization use financial analytics for now?Q |
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 24TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
FIGURE 2 | Use of Financial Analytics Within Three Years
Click on these icons to dig deeper.
What do you expect to be using financial analytics for within three years?Q |
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 25TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
FIGURE 3 | Types of Finance Data Drawn On for Analytics Activity Now
Click on these icons to dig deeper.
Which of the following types of finance-related data does your organization draw on now for analytics activity?Q |
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 26TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
Click on these icons to dig deeper.
FIGURE 4 | Types of Finance Data Drawn On for Analytics Within Three Years
Which of the following types of finance-related data do you expect your organization to draw on for analytics activity within three years?
Q |
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 27TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
Click on these icons to dig deeper.
FIGURE 5 | Use of Clinical Analytics Now
What does your organization use clinical analytics for now?Q |
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 28TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
Click on these icons to dig deeper.
FIGURE 6 | Use of Clinical Analytics Within Three Years
What do you expect to be using clinical analytics for within three years?Q |
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 29TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
Click on these icons to dig deeper.
FIGURE 7 | Types of Patient Data Drawn On for Analytics Now
Which of the following types of patient-related data does your organization draw on now for analytics activity?Q |
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 30TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
Click on these icons to dig deeper.
FIGURE 8 | Types of Patient Data Drawn On for Analytics Within Three Years
Which of the following types of patient-related data do you expect your organization to draw on for analytics activity within three years?
Q |
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 31TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
Click on these icons to dig deeper.
FIGURE 9 | Current Applications for Working With Large/Complex Data Sets
Which of the following best describes your current applications for working with large and/or complex data sets to reveal trends or specific insights?
Q |
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 32TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
Click on these icons to dig deeper.
FIGURE 10 | Presence of Downside Risk Contracts Prompting Need for Analytics Software
Has the presence of contracts with downside risk prompted the need for or increased dependence on analytics software or services? (Among those with downside risk)
Q |
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 33TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
FIGURE 11 | Financial Data Analytics Capabilities
How would you describe your financial data analytics capabilities?Q |
Click on these icons to dig deeper.
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 34TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
Click on these icons to dig deeper.
FIGURE 12 | Clinical Data Analytics Capabilities
How would you describe your clinical data analytics capabilities?Q |
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 35TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
Click on these icons to dig deeper.
FIGURE 13 | Top Data-Related Analytics Challenges Over Next Three Years
Please select the top three data-related challenges your organization expects to face in performing analytics over the next three years.
Q |
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 36TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
Click on these icons to dig deeper.
FIGURE 14 | Top Tactical Analytics Challenges Over Next Three Years
Please select the top three tactical challenges your organization expects to face in performing analytics over the next three years.Q |
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 37TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
Click on these icons to dig deeper.
FIGURE 15 | C-Suite Title Responsible for Financial Analytics
Which C-suite title within your organization is primarily responsible for your financial analytics activities?Q |
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 38TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
Click on these icons to dig deeper.
FIGURE 16 | C-Suite Title Responsible for Clinical Analytics
Which C-suite title within your organization is primarily responsible for your clinical analytics activities?Q |
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 39TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
QUESTIONS FOR YOUR TEAM
meeting guideTo address healthcare analytics issues, consider asking your leadership team these questions:
1. Do we recognize that the industry’s shift to delivering value-based care and accepting risk has increased the strategic importance of analytics, and that organizations that collect data from multiple sources and study data deeply will likely be more successful clinically and financially?
2. Do we understand the extent of the industry transformation that is coming, and that the trend involves both process transformation, mainly as it relates to care delivery, and information technology transformation, a component of which is analytics related?
3. Do we recognize that the need to measure, monitor, and be compensated for value-based care requires that analysis be available on a near-real-time basis? Are we preparing to take a longitudinal view of patient care activity, so that information systems can prompt caregivers when preventive action is likely to be needed?
4. Are we establishing the right platform and systems to document our performance along value-based lines, with the objective of being competitive as payers make network inclusion and reimbursement decisions?
5. Do we recognize the importance of data and interoperability to our organization, and are we ready to take a disciplined approach to developing strategies and tactics for the future? Are we prepared to update those strategies and tactics as required, while
at the same time maintaining a high degree of commitment to the direction?
6. Do we understand the value that comes from interoperability? Are we deploying workarounds to achieve near-term benefits of examining multiple sets of data while at the same time investigating and investing in longer-term, more robust solutions to work with data from multiple sources?
7. Do we recognize that achieving further improvements on cost containment and quality outcomes will rely on supporting our actions by obtaining data from multiple sources, examining data for causal relationships, and doing so in a way that is timely enough to prompt meaningful action?
8. Are we reviewing the skillsets of our talent pool, with the idea of identifying gaps in skills and filling them? Do we see the importance of developing analytics skills in the clinical team, and are we ensuring that providers understand the importance of documentation?
9. Do we recognize the need to balance the control and flexibility that in-house software development offers with the speed of implementation that comes with acquiring third-party software and outside services?
10. Have we established funding sources for the system development, computing resources, analytics software tools, and data and content experts necessary to achieve success?
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 40TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
Methodology
The 2016 Analytics in Healthcare Survey was conducted by the HealthLeaders Media Intelligence Unit, powered by the HealthLeaders Media Council. It is part of a series of monthly Thought Leadership Studies. In November 2015, an online survey was sent to the HealthLeaders Media Council and select members of the HealthLeaders Media audience. A total of 350 completed surveys are included in the analysis. The margin of error for a base of 350 is +/-5.2% at the 95% confidence interval.
Each figure presented in the report contains the following segmentation data: setting, number of beds (hospitals), number of sites (health systems), net patient revenue, and region. Please note cell sizes with a base size of fewer than 25 responses should be used with caution due to data instability.
ADVISORS FOR THIS INTELLIGENCE REPORTThe following healthcare leaders graciously provided guidance and insight in the creation of this report.
Stephen Beck, MD, FACP, FHIMSSChief Medical Informatics Officer Mercy HealthCincinnati, Ohio
Indranil “Neal” Ganguly, FCHIME, FHIMSS, CHCIOVice President and Chief Information OfficerJFK HealthEdison, New Jersey
Sue Schade, MBA, FCHIME, FHIMSSChief Information OfficerUniversity of Michigan Hospitals and Health Centers Ann Arbor, Michigan
UPCOMING INTELLIGENCE REPORT TOPICS
MARCH Payer/Provider Strategies
APRIL Mergers, Acquisitions, and Partnerships
MAY Emergency Department Strategies
C uncilHEALTHLEADERS MEDIA
Access. Insight. Analysis.Click for information on joining.
Click here to subscribe.
ABOUT THE HEALTHLEADERS MEDIA INTELLIGENCE UNITThe HealthLeaders Media Intelligence Unit, a division of HealthLeaders Media, is the premier source for executive healthcare business research. It provides analysis and forecasts through digital platforms, print publications, custom reports, white papers, conferences, roundtables, peer networking opportunities, and presentations for senior management.
Executive Vice President ELIZABETH PETERSEN epetersen@blr.com
Publisher CHRIS DRISCOLL cdriscoll@healthleadersmedia.com
Editorial Director BOB WERTZ bwertz@healthleadersmedia.com
Intelligence Unit Director ANN MACKAY amackay@healthleadersmedia.com
Custom Media Sales Operations Manager CATHLEEN LAVELLE clavelle@healthleadersmedia.com
Intelligence Report Contributing Editor SCOTT MACE smace@healthleadersmedia.com
Intelligence Report Design and Layout KEN NEWMAN
Intelligence Report Cover Art DOUG PONTE dponte@healthleadersmedia.com
Copyright ©2016 HealthLeaders Media, a division of BLR, 100 Winners Circle, Suite 300, Brentwood, TN 37027 Opinions expressed are not necessarily those of HealthLeaders Media. Mention of products and services does not constitute endorsement. Advice given is general, and readers should consult professional counsel for specific legal, ethical, or clinical questions.
Intelligence Report Senior Research Analyst MICHAEL ZEIS mzeis@healthleadersmedia.com
Intelligence Report Research Editor-Analyst JONATHAN BEES jbees@healthleadersmedia.com
FEBRUARY 2016 | The Analytics Challenge: Gaining Critical Insight Into Risk-Based Models PAGE 41TOC
THIS DOCUMENT IS PROTECTED. DO NOT COPY, SHARE, FORWARD, OR REPRODUCE.
Respondent Profile
Respondents represent titles from across the various functions at healthcare provider organizations.
Senior leaders | CEO, Administrator, Chief Operations Officer, Chief Medical Officer, Chief Financial Officer, Executive Dir., Partner, Board Member, Principal Owner, President, Chief of Staff, Chief Information Officer, Chief Nursing Officer, Chief Medical Information Officer
Clinical leaders | Chief of Cardiology, Chief of Neurology, Chief of Oncology, Chief of Orthopedics, Chief of Radiology, Dir. of Ambulatory Services, Dir. of Clinical Services, Dir. of Emergency Services, Dir. of Inpatient Services, Dir. of Intensive Care Services, Dir. of Nursing, Dir. of Rehabilitation Services, Service Line Director, Dir. of Surgical/Perioperative Services, Medical Director, VP Clinical Informatics, VP Clinical Quality, VP Clinical Services, VP Medical Affairs (Physician Mgmt/MD), VP Nursing
Operations leaders | Chief Compliance Officer, Chief Purchasing Officer, Asst. Administrator, Chief Counsel, Dir. of Patient Safety, Dir. of Purchasing, Dir. of Quality, Dir. of Safety, VP/Dir. Compliance, VP/Dir. Human Resources, VP/Dir. Operations/Administration, Other VP
Financial leaders | VP/Dir. Finance, HIM Director, Director of Case Management, Director of Patient Financial Services, Director of RAC, Director of Reimbursement, Director of Revenue Cycle
Marketing leaders | VP/Dir. Marketing/Sales, VP/Dir. Media Relations
Information leaders | Chief Technology Officer, VP/Dir. Technology/MIS/IT
Base = 121 (Hospitals)
Type of organization Number of beds
1–199 54%
200–499 21%
500+ 25%
Number of physicians
Base = 54 (Physician organizations)
1–9 24%
10–49 37%
50+ 39%
Region
WEST: Washington, Oregon, California,
Alaska, Hawaii, Arizona, Colorado, Idaho,
Montana, Nevada, New Mexico, Utah, Wyoming
MIDWEST: North Dakota, South Dakota,
Nebraska, Kansas, Missouri, Iowa, Minnesota,
Illinois, Indiana, Michigan, Ohio, Wisconsin
SOUTH: Texas, Oklahoma, Arkansas,
Louisiana, Mississippi, Alabama, Tennessee,
Kentucky, Florida, Georgia, South Carolina,
North Carolina, Virginia, West Virginia, D.C.,
Maryland, Delaware
NORTHEAST: Pennsylvania, New York,
New Jersey, Connecticut, Vermont, Rhode
Island, Massachusetts, New Hampshire, Maine
Title
Base = 350
51%Senior leaders
6% Marketing
leaders
0
10
20
30
40
50
60
17% Clinicalleaders
18% Operations
leaders
6% Financial leaders
35%
27%
19%
20%
Number of sites
Base = 100 (Health systems)
1–5 12%
6–20 28%
21+ 60%
Base = 350
Hospital 35%
Health system (IDN/IDS) 29%
Physician organizations 15%
Long-term care/SNF 9%
Health plan/insurer 5%
Ancillary, allied provider 4%
Government, education/academic 3%
2% Information
leaders
➔
Join today at www.healthleadersmediacouncil.com
Be a voiceGain insight from your peersShape the direction of the industry
lC unci The nation’s most exclusive
healthcare intelligence community
Recommended