This white paper authored by Jason Oliveira discusses the marriage between business and clinical decision support systems within the healthcare industry.
3. FOREWORD
A ROSE BY ANY OTHER NAME MAY STILL BE A ROSE, BUT DECISION-SUPPORT
SYSTEMS BY ANY OTHER NAME ARE MOST LIKELY KISSING COUSINS AT BEST.
ON ONE SIDE OF THE FAMILY FENCE ARE THE VARIOUS BUSINESS DECISION
SUPPORT SYSTEMS THAT SUPPORT BUDGETING, EXECUTIVE DECISION-MAKING,
FINANCIAL ANALYSIS, QUALITY MANAGEMENT, AND STRATEGIC PLANNING, TO
NAME BUT A FEW. On the other side of the fence are the evolving clinical decision support
systems that support results reporting, pharmaceutical ordering and dispensation, differential
diagnoses, real-time clinical pathways, dynamic literature research, and clinical alerts. These
two types of decision support systems, business and clinical, differ significantly both in intent
and content, but are all too often incorrectly referenced interchangeably.
The purpose of this paper is twofold. The first is to define the high-level delineation
of the very different intent, content, and methods of the business and clinical decision-making
functions. As form must follow function, the information technology and methodologies for
data modeling, management, and presentation to the user also differ for business and clinical
decision-making. While the intent, content, and methods may differ, there are still many common
elements of the two decision-support approaches that can be shared to great benefit.
The concepts of commonality and sharing lead us into the second purpose of this paper.
When clinical decision support (CDS) is integrated with business decision support (BDS), a
marriage occurs that is mutually beneficial. This marriage is not an easy matter, and will
occur and succeed only under pressure of well-planned integrated data and decision support
system strategies.
A SHOTGUN WEDDING | BUSINESS DECISION SUPPORT MEETS CLINICAL DECISION SUPPORT
4. THIS ARTICLE WAS ORIGINALLY PUBLISHED IN THE FALL 2002 ISSUE OF THE
JOURNAL OF HEALTHCARE INFORMATION MANAGEMENT PUBLISHED BY THE
HEALTHCARE INFORMATION AND MANAGEMENT SYSTEMS SOCIETY. IT IS
REPRINTED WITH PERMISSION.
ABOUT THE AUTHOR
JASON OLIVEIRA, MBA, IS A HEALTHCARE DECISION SUPPORT SPECIALIST AND
MANAGER WITH THE HEALTH CARE CONSULTING GROUP OF KURT SALMON
A S S O C I AT E S . FOR MORE I N F O R M AT I O N , PLEASE EMAIL JA S O N AT
JOLIVEIRA@KURTSALMON.COM.
KURT SALMON ASSOCIATES | INSIGHTS, JANUARY 2003
5. CONTENTS
A SHOTGUN WEDDING: BUSINESS DECISION SUPPORT MEETS CLINICAL DECISION SUPPORT
T H E C H A N G E I N O R G A N I Z AT I O N A L C U LT U R E A N D T H E R E D E S I G N O F B U S I N E S S
AND CLINICAL PROCESSES THAT ALLOW THE USE OF EMPOWERED DECISION-MAKING
TOOLS IS BY FAR THE MORE DIFFICULT TASK FACING INFORMATION TECHNOLOGISTS
A N D O R G A N I Z AT I O N A L L E A D E R S . K S A T H O U G H T P I E C E , “A S H OT G U N W E D D I N G :
B U S I N E S S D E C I S I O N S U P P O R T M E E T S C L I N I C A L D E C I S I O N S U P P O R T,” O F F E R S A
G L I M P S E O F H OW A G R E AT B E N E F I T O C C U R S D E S P I T E T H E V E RY D I F F E R E N T
I N T E N T, C O N T E N T, A N D M E T H O D S O F T H E B U S I N E S S A N D C L I N I C A L D E C I S I O N -
M A K I N G F U N C T I O N S . By effectively closing the loop between the data, analytics,
processes, and methods supporting business and clinical decision-making, a health care
organization closes the loop between its knowledge generation activities and its actions
at the bedside: knowledge guiding actions, actions generating knowledge.
OVERVIEW
1 SECTION ONE
A DECISION-MAKING MODEL
The core objective of both clinical and business decision-support systems is to enhance
a decision-making process. Their differences are clear, yet widely misunderstood.
5 SECTION TWO
THE DECISION-SUPPORT ARCHITECTURE
Bringing the clinical and business decision-making processes together requires a
sophisticated architecture. This architecture must include acquisition, organization and data
exploitation functions. These functions are driven by Computer-based Patient Record,
data warehousing, and the clinical data repository.
13 SECTION THREE
LET’S HAVE A SHOTGUN WEDDING
Various data exploitation tools deployed to decision makers produce a decision loop of the
business and clinical decision-support systems creating improvements in the decision-
making process.
7. A DECISION
MAKING MODEL
AS THE GOAL OF BOTH BDS AND CDS IS TO ENHANCE A DECISION-
1
MAKING PROCESS, A MODEL OF THAT PROCESS WILL FACILITATE THE
DISCUSSION OF THEIR DIFFERENCES. AT ITS MOST BASIC LEVEL, A
DECISION IS A CHOICE BETWEEN ALTERNATIVE COURSES OF ACTION DEALING
WITH AN ISSUE.
THE DECISION-MAKING MODEL IS COM- issue. Finally, in order to improve the quality
PRISED OF FIVE STEPS: 1) INTELLIGENCE of future decisions, evaluate the results of
The heart of decision-
GATHERING, 2) DEVISING SOLUTION ALTER- the decision to assess how well it
making is to then devise,
NATIVES, 3) CHOOSING THE BEST SOLU- addressed the issue at hand.
evaluate, and choose
TION, 4) IMPLEMENTING THE SOLUTION, This decision-making model remains
from numerous alternative
AND 5) EVALUATING ITS EFFECTIVENESS. the same whether you are deciding whether
solutions the one that
Intelligence gathering denotes situa- to acquire a community hospital (i.e., BDS),
best addresses the formu-
tional fact finding in order to better define or you are deciding on the best therapeutic
lated issue.
what is happening. The description of what regimen for a cancer patient (i.e., CDS).
is happening will coalesce into the design What does differ between the two types of
of a concrete issue that requires one or decisions under discussion are the charac-
more decisions to be made. The heart of teristics of their decision-making processes
decision-making is to then devise, evaluate, within the model. These differing process
and choose from numerous alternative characteristics include temporal use, goal-
solutions the one that best addresses the orientation, and the level of structure
formulated issue. With presumably the best involved in business and clinical decision-
alternative solution in hand, then implement making.
it in the hopes to positively address the
A SHOTGUN WEDDING | BUSINESS DECISION SUPPORT MEETS CLINICAL DECISION SUPPORT
8. THE CLINICAL DECISION- with the desirability of a business goal differ
MAKING PROCESS from those of the clinician presented with a
The typical decision-making sick patient. For the purposes of this paper,
clinical
process, not the overarching decision-making a healthcare business strategist encom-
model, differs significantly from the typical passes all decision processes other than
business decision-making process. Clinical direct patient-care delivery, even if clinical in
decisions by their nature are real-time and nature. These business decision processes
are often performed at the point of care. A include strategic planning, budgeting, and
patient presents indications, and a series financial analysis as well as, quality man-
of decisions need to be made now to save agement programs, clinical process
a life, alleviate the symptoms, and cure the improvement, and clinical benchmarking.
underlying disease/condition. Business decision-making can occur at a
Clinical decisions are specifically strategic, tactical, and operational level. Bayesian probability
goal oriented, that is, a cure and/or the This paper addresses the information and strengthens the determi-
alleviation of symptoms are sought. The system needs of strategic and tactical deci- nation of a disease as
clinical goal is first reached through intelli- sion-making only. Applications supporting the symptoms that are
gence gathering and the making of a diag- daily operational decisions and processes most probably present
nosis, often using Bayesian probability. such as ADT, registration, scheduling, and given a disease are deter-
2
Simply stated, Bayesian probability patient billing are not included in our con- mined through evaluation
strengthens the determination of a disease sideration. Clinical decisions, by definition, and diagnostic testing.1
as the symptoms that are most probably are operational in nature.
present given a disease are determined What remains for our consideration
through evaluation and diagnostic testing.1 are business decisions that are batch ori-
The clinician approximates the probabilities ented in nature. That is, business problems
of symptom/sign and disease combina- which are addressed occasionally and not
tions through an understanding of the real-time. The business decision is not
underlying physiology, experience with pre- concerned with a singular element, such as
vious similar cases, and literature review of a sick patient, but large aggregations of
similar cases. An intelligence gathering many elements that address an ill-defined
process called case-based reasoning. problem such as how can costs be
The clinical decision-making process reduced, or clinical outcomes improved.
can be characterized as being very structured The aggregation of data is largely for the
and goal oriented within a real-time clinical purposes of intelligence gathering, as
context, which is significantly different then opposed to the purposes of addressing an
the characterization of the business decision- already known specific goal. These charac-
making process. teristics make the business decision-making
process unstructured, goal searching, and
THE BUSINESS DECISION- long range in nature. Figure 1 summarizes
MAKING PROCESS the characteristics of the business and
The issues and problem solving process of clinical decision-making process.
a health care business strategist presented
KURT SALMON ASSOCIATES | INSIGHTS, JANUARY 2003
9. FIGURE 1
Clinical decisions, by defi- DECISION-MAKING PROCESS CHARACTERISTICS
nition, are operational in
nature. CHARACTERISTIC BUSINESS CLINICAL
Retrospective, batch, Real-time, case based,
s s
TEMPORAL USE
long-range operational
Unspecified intelligence Specified goal seeking
s s
GOAL ORIENTATION
gathering, goal searching
oriented
Unstructured Very structured,
s s
STRUCTURE OF DECISION
Bayesian
Source: KSA Analysis
3
A SHOTGUN WEDDING | BUSINESS DECISION SUPPORT MEETS CLINICAL DECISION SUPPORT
10. 4
KURT SALMON ASSOCIATES | INSIGHTS, JANUARY 2003
11. THE DECISION-SUPPORT
ARCHITECTURE
INTEGRATING AND MANAGING THE CLINICAL AND BUSINESS DECISION-
5
MAKING PROCESSES OF A HIGHLY DIVERGENT REPUBLIC OF PROFESSIONAL
DISCIPLINES REPRESENTED IN EVEN THE SMALLEST OF HEALTH CARE
ORGANIZATIONS REQUIRES A ROBUST AND SOPHISTICATED DECISION-
SUPPORT ARCHITECTURE. 2 AT ITS MOST BASIC LEVEL, EMPOWERING THE CLINICAL
AND BUSINESS DECISION-MAKERS OF THE ORGANIZATION INVOLVES THE
ACQUISITION, ORGANIZATION AND EXPLOITATION OF HIGH QUALITY INFORMATION AT
THE RIGHT TIME, THROUGH THE RIGHT MEDIUM, AND TO THE RIGHT DECISION-MAKER.
THE FUNCTIONALITY DELIVERED BY THE ORGANIZATION. The ability to efficiently
s
At its most basic level, DECISION-SUPPORT ARCHITECTURE, AS model, store and retrieve the data with
empowering the clinical DEPICTED IN FIGURE 2, SHOULD INCLUDE applied business and clinical rules and
and business decision- T H E F O L L OW I N G : semantics at both a logical data model
ACQUISITION. The means to acquire data
makers of the organization s and physical database layer.
from the numerous internal operational
involves the acquisition, The various retrieval,
s EXPLOITATION.
information systems supporting the real-
organization and exploita- reporting, analysis and decision support
time clinical, financial and administrative
tion of high quality infor- tools used to derive and deliver informa-
processes. Also to be included is the
mation at the right time, tion from the acquired and organized data.
acquisition and integration of external data
through the right medium, These three functions have been
sources such as supplied by data vendors,
and the right decision- addressed by various health care informa-
state and federal based data agencies,
maker. tion technology initiatives. These initiatives
best-practice sources, and by the organiza- include the Computer-based Patient Record
tion’s external business partners. (CPR)3, data warehousing, and the clinical
A SHOTGUN WEDDING | BUSINESS DECISION SUPPORT MEETS CLINICAL DECISION SUPPORT
12. data repository (CDR). The CPR is an logical separation of a health care institu-
over-arching vision that includes all the tion’s operational data systems and its ret- A data warehouse is, simply
elements of capturing, storing, processing, rospective analytical decision-support activi- stated, the physical and
communicating, and presenting patient ties. The fundamental requirements of the logical separation of a
record information and related data and operational and analytical decision support healthcare institution’s
knowledge bases. Supporting the data systems are very different. The operational operational data systems
and knowledge base of the CPR vision information systems need peak perform- and its retrospective ana-
are both data warehousing and the clinical ance for a set of small structured real-time lytical decision-support
data repository. transactions. Whereas, the analytical deci- activities.
sion support system needs flexibility and
T H E DATA WA R E H O U S E broad scope for yet to be defined retro-
Data warehousing is an old concept that spective analytical needs. It is undesirable
has taken on new strategic implications to have retrospective analysis interfere with
within the health care industry. A data ware- and degrade the performance of opera-
house is, simply stated, the physical and tional systems. The primary concept of
FIGURE 2
6 DECISION-SUPPORT ARCHITECTURE
ACQUISITION O R G A N I Z AT I O N E X P L O I TAT I O N
Outcomes Planning/
s
marketing
Data Research
s
ols
warehouse
m
To
for
Planning
Support
& Trans
Performance
s
Internal
evaluation
and
external
act
ion
systems Outcome/
s
xtr
cis
disease
Clinical
E De
Finance management
data
repository
Finance plus
s
more
Quality
Indicators
Source: KSA Analysis
KURT SALMON ASSOCIATES | INSIGHTS, JANUARY 2003
13. data warehousing is to most effectively identification of all Medicare patients in a
access data stored for business and clinical health network for the past ten years no
analysis by separating and integrating it matter which of dozens of Medicare insur-
from the data in numerous internal and ance codes were used in five different oper-
external operational information systems. ational billing systems.
Data warehouses are most successful As evidenced above, business data
when data is integrated from more than analysis has a need for a huge breadth and
one operational system as well as with depth of data — and not just data, but
external market, benchmarking, and com- information. Turning data into information
petitor data sources. Another key attribute involves reorganizing operational data,
of the data in a data warehouse is that it deriving new data, integrating disparate
has become mostly non-volatile. This data, and delivering information to busi-
means that after the data is loaded into the ness decision-makers through various data
data warehouse, there are little to no mod- exploitation tools to be discussed later.
ifications made to this information. While Conversely, the needs of clinicians delivering
an ICU monitoring system, an operational real-time clinical care to patients require
clinical system, can capture and trend structured, defined, goal-oriented support
blood pressure readings continuously, it from clinical decision-support systems. As
7
would only be desirable to capture, for busi- the business decision-support system is
ness analysis, the admission and dis- built on the informational foundation of a
charge BP measures of a patient. data warehouse, so is the clinical decision
The two remaining key attributes of a support system built on the foundation of a
Data warehouses are
data warehouse are its logical and physical clinical data repository.
most successful when
data models. The warehouse logical data
data is integrated from
T H E C L I N I C A L DATA
model aligns with the analytical, versus
more than one opera-
REPOSITORY
operational, data needs of the health
tional system as well as
Clinical professionals, information officers,
organization. The data entities defined and
with external market,
and medical informaticians have differing
maintained in the data warehouse parallel
benchmarking, and com-
notions of what clinical data repositories
analytical entities such as product lines,
petitor data sources.
should do and how they differ from other
catchment areas, clinical services, provider
types of databases, namely the business
groups, referral sources, costs, and profits.
data warehouse described above. Depending
This is as compared to operational data
on whether health care organizations are
models that contain entities designed for
trying to support real-time clinician needs or
processes such as charge posting, ordering,
strategic and research objectives, two very
resulting, discharging, and cash posting.
different types of databases are required.
At a physical level, the warehouse is
Clinical data repositories are designed for
designed to efficiently deliver information for
immediacy and support real-time, struc-
analytical purposes, versus operational trans-
tured, integrated clinical decision-support.
action processing purposes. This efficiency is
Data warehouses are designed to support
gained through the use of several techniques,
batch, retrospective, and unstructured busi-
among which include the de-normalization,
ness decision support, including clinically
aggregation, hyper-indexing, and standardiza-
oriented business decisions.4 All to often
tion of data. These data transformation tech-
what is really a data warehouse is
niques allow, for example, the simple and fast
A SHOTGUN WEDDING | BUSINESS DECISION SUPPORT MEETS CLINICAL DECISION SUPPORT
14. described as a clinical data repository, or agement technologies. However, the two data
vice versa, a clinical data repository is management technologies are designed to
claimed to be able to effectively support support two very different decision-making
analytical decision making. processes. The acquired and organized
A CDR is a complementary technology data now needs to be exploited by decision-
for a Computer-based Patient Record. The makers through the use of software tools The CDR is designed to
CDR can be viewed as a patient-focused and methods that transform the data into provide the clinical view of
clinical data store for the CPR. The CDR is actionable information. a patient to a clinician in
designed to provide the clinical view of a real-time to support clinical
DATA E X P L O I TAT I O N
patient to a clinician in real-time to support decision-making.
Data exploitation refers to the various data
clinical decision-making. The CDR consoli-
retrieval, reporting, decision support, and
dates and integrates the disparate sources
analysis tools used to derive and deliver
of operational clinical data that reside in
information from the acquired and organ-
laboratory, radiology, ambulatory care,
ized data in the data warehouse and the
dietary, and numerous other clinical infor-
clinical data repository. These data
mation systems. Presenting to the clinician
exploitation tools are the means through
at the bedside, the whole clinical picture of
which business analysts, operational man-
the patient under their care.
8
agers, and clinicians view, integrate, and
The field of medical informatics fos-
analyze the various data stores that have
tered in the world’s academic medical cen-
been discussed above. It is through the
ters is creating the infrastructure to realize
tools that data is transformed into action-
the CDR, and through its application, the
able information through targeted subject
CPR. The CDR is the culmination of years of
specific algorithms, analysis, measure-
research developing the components
ment, summarization, reports, and specific
required to build it. These components
decision-support logic.
include the structured medical vocabulary
QUERIES. Queries are the basic mecha-
systems such as ICD-9-CM, CPT4, s
nism, typically using the Structured
SNOMED, Arden Syntax, Medical Logic
Query Language (SQL), to efficiently
Modules, and LOINC. The components also
search and retrieve detail data from the
include the basic mechanisms of data inter-
two organized data stores. The CDR is
change, which include CORBAMed, HL7,
optimized to answer queries that retrieve
DICOM, and ASTM protocols. Last, but likely
the clinical data of a single patient. The
to be the most difficult to achieve, is the
data warehouse is optimized to answer
standardization of encoding and represent-
queries that retrieve the data for thou-
ing medical knowledge itself, such as the
sands of patients over numerous years.
Intermed Common Model and Guideline
REPORTS. Reporting is the ubiquitous
Interchange Format (GLIF). s
tool of displaying detail and summarized
The data warehouse and the clinical
data both online and through printing.
data repository are, at their core, data man-
KURT SALMON ASSOCIATES | INSIGHTS, JANUARY 2003
15. Reporting tools are typically integrated its name from the imagery of having to
with query tools. The later retrieves the dig through gigabytes or terabytes of
data, the former summarizes, formats, ‘rock’ (i.e., raw data) to find that small
and displays the data to the user. nugget of actionable information ‘gold’.
The combination of modeling techniques
s ON-LINE ANALYTICAL PROCESSING (OLAP).
On-Line Analytical Processing includes enables the discovery of relationships,
those tools that summarize data in pre- patterns, trends, and predictive models in
determined manners to allow the effi- the data warehouse and clinical date
cient navigation of that data during a repository not easily found through tradi-
free-form data analysis session. This tional decision-support tools.
capability is most commonly associated s DECISION SUPPORT SYSTEM (DSS).
Data mining derives its
with multi-dimensional data cubes where Those routine decisions that are struc-
name from the imagery of
data is summarized into analytical tured enough can be embodied in the
having to dig through giga-
dimensions such as fiscal period, cost logic of a targeted decision support sys-
bytes or terabytes of
center, corporate division, budgeted and tem. Examples of these certainly include
‘rock’ (i.e., raw data) to
actual expenses. The OLAP tool then diagnosis expert systems, clinical alerts,
find that small nugget of
allows the user to quickly and easily ‘drill- and assisted prescription ordering on the
actionable information
down’ between the data dimensions at clinical decision-support systems end.
‘gold’.
9
any level of summarization, from corporate Business decision support systems
overview down to the cost center level. include clinical pathway development,
DATA MINING. Data mining is the collec- enterprise resource management, budg-
s
tive term of the numerous techniques eting, strategic planning, and cost
and methodologies that have found their accounting systems. Decision-support
origin in several fields of study including systems are usually comprised of the
artificial intelligence, machine learning, query, reporting, OLAP and data mining
,
pattern recognition, advance statistical technologies described above. These
modeling, and data visualization. These technologies are in a sense the develop-
fields of study have coalesced from theory ment components for an application
into the targeted application of modeling designed to support a specific set of
techniques to the discovery of knowledge decision-making processes.
in large databases. Data mining derives
A SHOTGUN WEDDING | BUSINESS DECISION SUPPORT MEETS CLINICAL DECISION SUPPORT
16. 10
KURT SALMON ASSOCIATES | INSIGHTS, JANUARY 2003
17. LET’S HAVE A
SHOTGUN WEDDING
THE DATA WAREHOUSE, THE CLINICAL DATA REPOSITORY, AND THE SET
11
OF DATA EXPLOITATION TOOLS ARE COMPLEMENTARY INFORMATION
TECHNOLOGIES EACH DESIGNED FOR DIFFERENT DECISION-SUPPORT
NEEDS. SOME ARE FOR RETROSPECTIVE FINANCIAL, CLINICAL, AND
OPERATIONAL BUSINESS ANALYSIS. Some are for real-time, integrated delivery of
patient-centric clinical data and medical knowledge to the clinician.
WHILE CONTENT AND INTENT MAY DIFFER, CLOSED-LOOP DECISION MAKING
The decision loop refers THERE ARE COMMON ELEMENTS OF THE The marriage of business and clinical
to the fact that decisions TWO DATA AND SYSTEM STRATEGIES THAT decision support is realized through a
CAN BE SHARED. Non-volatile, historical
as recorded in a clinical decision loop that is made evident in the
clinical data from a CDR can feed a data
decision support system various data exploitation tools deployed
warehouse to support an OLAP clinical-
can feed a business deci- to decision-makers, both business and
pathway utilization tool. Cost data from a
sion support system. clinical. The decision loop refers to the
data warehouse can feed a CDR to support fact that decisions as recorded in a clinical
a cost-effectiveness driven case management decision-support system can feed a busi-
decision-support system. The remaining ness decision-support system. The deci-
section of this paper highlights the synergies sions as recorded in a business decision-
that can be realized from well-planned, inte- support system then, in turn, can feed the
grated data store and data exploitation clinical system. The decision loop creates
strategies. improvements in the decisions made on
both sides of the decision process fence.
A SHOTGUN WEDDING | BUSINESS DECISION SUPPORT MEETS CLINICAL DECISION SUPPORT
18. There are numerous examples of decision information systems across multiple settings
loops that would benefit from integrated data of care (i.e., hospitals, physician offices, Managing the effective-
and application strategies. The decision loop nursing home, patient’s home), case man- ness of a case manage-
of case management will be discussed in agement is hastening the development of ment strategy requires the
detail. Additional decision loops would linkages between these fragmented data development of significant
include outcomes management, strategic sources into the clinical data repository dis- and effective care plans
planning, benefits management, capitation cussed above. Managing the effectiveness and measuring compliance
management, disease management, and of a case management strategy requires to those plans.
contract modeling to name but a few. the development of significant and effective
care plans and measuring compliance to
CASE MANAGEMENT those plans. The data warehouse is in the
Because case management requires timely best position to support the analysis of
access to patient data that is currently case management effectiveness across
collected and stored in many different multiple clinical services, providers, and
places by many different operational clinical patients.
FIGURE 3
12 CLOSED LOOPED DECISION MAKING FOR CASE MANAGEMENT
A n al
yt ic
al in
Strategic decisions fo r m
s Identify high cost a ti o
nu
populations se
s Compare against
regional best practice
benchmarks
s Choose a high volume
population with a high
Tactical decisions
variance
s Critical pathway
development
s Best practice resource
utilization profile
s Variance reporting
s Physician reporting
Operational decisions
s Critical alerts
s Critical pathway enabled
order entry/results
s Approved formularies
at prescription
s Dynamic literature
searches
Source: KSA Analysis
KURT SALMON ASSOCIATES | INSIGHTS, JANUARY 2003
19. The decision loop for case manage- CONCLUSION
ment, as depicted on page 12 in Figure 3, As we always advise, data and application
starts at the identification of a patient strategies are only a collection of tools, it is
group for whom the application of case essential that the health care organization
management will result in significant is prepared to take advantage of them. The
improvements in clinical and cost effective- change in organizational culture and the
ness. Data mining tools can apply statisti- redesign of business and clinical processes
cal clustering techniques against the data that allow the use of empowered decision
warehouse to determine categories of making tools is by far the more difficult
patients that have similar clinical indica- task facing information technologists and
tions and high costs.5 The source of the organizational leaders. A firm understanding
clinical data being the clinical data reposi- of business improvement methods, corpo-
tory, and of the patient costs being the data rate business and clinical goals, and the
warehouse. Statistical regression tools of information strategies themselves is a
data mining can then identify which clinical requirement to realize significant benefits.
factors are most highly correlated to high But most importantly, the realization that
costs. Patient age, high-blood pressure, clinical and business processes are not
and pharmaceutical utilization being exam- mutually exclusive, therefore, neither are
13
ples. This data then can be used to a devel- their decision-support strategies.
op a cost-effective clinical pathway for this At no other time in the history of the
patient group. health care industry have market impera-
The clinical pathway is deployed tives demanded the marriage of business
The future of information through a clinical decision support system and clinical decision support. Clinical out-
technology and its inte- used by both clinicians and case managers. comes research and the care delivery
grated application to both The real-time clinical data needs of the process were clearly the domain of white-
sides of the decision- pathway are supported by the clinical data coated clinicians. Cost cutting and reim-
support fence will serve repository. Furthermore, the pathway can bursement maximization were clearly the
as the proverbial shotgun be integrated with the organization’s opera- purview of business-suited MBAs and
to bring these two disci- tional Order Entry and Results Reporting CPAs. The future of information technology
plines together in marital application to ensure pathway suggestions and its integrated application to both sides
bliss. of lab tests and approved formularies are of the decision-support fence will serve as
adhered to at the point-of-care. The meas- the proverbial shotgun to bring these two
urement of costs, clinical outcomes, and disciplines together in marital bliss. This
quality as captured by those respective marriage will not be an easy matter. It will
decision support systems are fed back to require a lot of marriage counseling on part
the data warehouse, and now available for of information technologists and enlight-
aggregated clinical pathway utilization and ened health organization leaders, but the
cost-effectiveness analysis using OLAP and result will be years of financial health and
reporting tools. The decision loop is closed clinical care improvements.
as new clinical pathways are created and
existing ones improved at the retrospective
business decision-support level, and
deployed at the real-time operational clinical
decision-support level.
A SHOTGUN WEDDING | BUSINESS DECISION SUPPORT MEETS CLINICAL DECISION SUPPORT
20. FOOTNOTES
1
START, State of the Art: Oncology in Europe.
www.oncoweb.com/start/chapt-05/chap5-2.htm. Section 2. Decision Theory, 1998, p. 5.
2
Oliveira, J.D., and Lederman M. Decision Support and Executive Information Systems.
Advance for Healthcare Information Executives, August 1998, p. 46.
3
Dick, R.S., and Steen, E.B. (Eds.). The Computer-based Patient Record: An Essential
Technology for Health Care. Washington, DC: National Academy Press, 1991.
4
Morrisey, John. Differing perceptions about CDRs complicate purchases, impede
advances. Modern Healthcare, October 1998, p. 57.
5
Oliveira J.D., Mining for Information Gold: Data Mining and its Healthcare Application.
Advance for Healthcare Information Executives, January 1999.
14
KURT SALMON ASSOCIATES | INSIGHTS, JANUARY 2003