Demystifying Healthcare Data Governance 2014

© 2014 Health Catalyst
www.healthcatalyst.com
Creative Commons Copyright
www.healthcatalyst.comCreative Commons Copyright
Dales Sanders – May 7, 2014
Demystifying Healthcare Data
Governance

Today’s Agenda
 General concepts in data governance
 Unique aspects of data governance in
healthcare
 The layers and roles in data governance
 Constant theme: Data governance as it relates
to analytics and data warehousing
2

A Sampling of My Up & Down Journey
TOO LITTLE DATA
GOVERNANCE
TOO MUCH DATA
GOVERNANCE
WWMCCS: Worldwide Military Command & Control System
MMICS: Maintenance Management Information Collection System
NSA: National Security Agency
IMDB: Integrated Minuteman Data Base
PIRS: Peacekeeper Information Retrieval System
EDW: Enterprise Data Warehouse
(1986)
WWMCCS
(1987)
MMICS
(1992)
NSA Threat
Reporting
(1995)
IMDB
& PIRS
(1996)
Intel
Logistics
EDW
(1998)
Intermountain
Healthcare
(2005)
Northwestern
EDW
(2009)
Cayman
Islands HSA
1983
2014
3

The Sanders Philosophy of
Data Governance
The best data governance governs
to the least extent necessary to
achieve the greatest common good.”
Govern no data until its time.”
4

Centralized EDW;
monolithic early
binding data model
Data Governance Cultures
HIGHLY
CENTRALIZED
GOVERNMENT
BALANCED
GOVERNMENT
HIGHLY
DECENTRALIZED
GOVERNMENT
Centralized EDW;
distributed late
binding data model
No EDW; multiple,
distributed analytic
systems
5

Characteristics of Democracy
 Elements of centralized decision making
● Elected or appointed, centralized representatives
● Majority rules
 Elements of decentralized action
● Direct voting and participation, locally
● Everyone is expected to participate in developing
shared values, rules, and laws; then abide by them
and act accordingly
6

What’s It Look Like?
Not enough data governance
 Completely decentralized, uncoordinated data analysis
resources-- human and technology
 Inconsistent analytic results from different sources,
attempting to answer the same question
 Poor data quality, e.g., duplicate patient records rate is >
10% in the master patient index
 When data quality problems are surfaced, there is no formal
body nor process for fixing those problems
 Inability to respond to new analytic use cases and
requirements… like accountable care
7

What’s It Look Like?
Too much data governance
 Unhappy data analysts… and their customers
 Everything takes too long
– Loading new data
– Making changes to data models to support new analytic use cases
– Getting access to data
– Resolving data quality problems
– Developing new reports and analyses
8

Poll Question
What best describes the current state of affairs for
data governance in your organization?
193 Respondents
Authoritarian – 19.7%
Democratic – 24.3%
Tribal – 56%
9

Poll Question
How would you rate data governance effectiveness
in your organization?
179 Respondents
5 – Very effective – 1.6%
4 – 7.2%
3 – 22.3%
2 – 44.1%
1 – Ineffective – 24.8%
10

The Triple Aim of Data Governance
1. Ensuring Data Quality
• Data Quality = Completeness x Validity
2. Building Data Literacy in the organization
• Hiring and training to become a data driven company
3. Maximizing Data Exploitation for the
organization’s benefit
• Pushing the data-driven agenda for cost reduction,
quality improvement, and risk reduction
11

Keys to Analytic Success
The Data Governance Committee should be a
driving force in all three…
– Setting the tone of “data driven” for the culture
– Actively building and recruiting for data literacy
among employees
– Choosing the right kind of tools to support
analytics and data governance
Mindset
Skillset
Toolset
12

The Data Governance Layers
Happy Data
Analyst
13

The Different Roles in Each Layer
Executive & Board Leadership
We need a longitudinal analytic view across the
ACO of a patient’s treatment and costs, as well
as all similar patients in the population we serve.”
14

Data Governance Committee
We need an enterprise data warehouse
that contains all of the clinical data and
financial data in the ACO, as well as a
master patient identifier.”
We need a data analysis team, as well as
the IT skills to manage a data warehouse.”
The following roles in the organization
should have the following types of access
to the EDW.”
15

Data Stewards
I’m responsible for patient
registration. I can help.”
I’m responsible for clinical
documentation in Epic. I can help.”
I’m responsible for revenue cycle
and cost accounting. I can help.”
16

Data Architects & Programmers
We will extract and organize the data from the
registration, EMR, rev cycle, and cost
accounting and load it into the EDW.”
“Data stewards, can we sit down with you and
talk about the data content in your areas?”
“DBAs and Sys Admins, here are the roles
and access control procedures for this data.”
17

DBAs & System Administrators
Here is the access control list and
procedures for approving access to this
data. Let’s build the data base roles and
audit trails to support these.”
18

Data access & control system
When this person logs in, they have the
following rights to create, read, update,
and delete this data in the EDW.”
19

Data Analysts
I’ll log into the EDW and build a query
against the data in the EDW that should be
able to answer these types of questions.”
“Data Stewards, can I cross check my
results with you to make sure I’m pulling
the data properly?”
“Data architects, I’ll let you know if I have
any trouble with the way the data is
organized or modeled.”
20

Who Is On The Data Governance
Committee?
Representing the
analytics customers
The data technologist
The clinical data owners
The financial and supply
chain data owner
Representing the
researchers’ data needs
Chief Analytics Officer
CIO
CMO & CNO
CFO
CRO
CMIO
21
The clinical data
technologist

Data Governance Committee Failure Modes
Wandering: Lacking direction and experience
● “We know we need data governance, but we don’t know how to go about it.”
Technical Overkill: An overly passionate and inexperienced IT person leads the
data governance committee
● Can’t see the forest for the trees
● For example, Executives on the Data Governance Committee (DGC) are asked
to define naming conventions and data types for a database column
Politics: Members of the DGC are passive aggressive, narrowly motivated, data poseurs
● They pretend to be data driven and selfless, but they aren’t
● Territorial and defensive about “their” data
● “That person isn’t smart enough to use my data properly.”
Red Tape: Committee members are not governors of the data, they are bureaucrats
● Red tape processes for accessing data
● Confuse data governance with data security
22

Poll Question
Your organization’s biggest risks to the success of
the Data Governance Committee
182 Respondents – Multiple Choice
Wandering – 52%
Politics – 61%
Technical Overkill – 20%
Red Tape – 36%
Other – 16%
23

Data Governance & Data Security
 Data Governance Committee: Constantly pulling for broader
data access and more data transparency
 Information Security Committee: Constantly pulling for
narrower data access and more data protection
 Ideally, there is overlapping membership that helps with the
balance
24

Tools for Data Governance
Data quality reports
– Data Quality = Validity x Completeness
CRM tools for the data warehouse
– Who’s using what data? When? Why?
“White Space” data management tools
– For capturing and filling-in computable data that’s missing in the
source systems
Metadata repository
– What’s in the data warehouse?
– Are there any data quality problems?
– Who’s the data steward?
– How much data is available and over what period of time?
– What’s the source of the data?
25

Practice
Protocols
Processing
EDW
Analyzable data
Clinicians use diverse
protocols & orders in
daily care
Sub-Optimal State
The Four Levels of Closed Loop Analytics in Healthcare
© 2014 Denis Protti, Dale Sanders & Corinne Eggert
CDS:
EDW:
EHR:
MTTI:
Clinical Decision Support
Enterprise Data Warehouse
Electronic Health Record
Mean Time To Improvement
Clinical Information
Systems
Decisions & Actions
Supporting information
Clinical, EHR, EDW &
Analytics Teams
Align metrics & data
Update EHR & EDW
with new data items if
needed & possible
Start here
Monitor baselines &
clinical processes
Select a problem
Set outcomes & metrics
Quality
Governance
Clinical Variations
& Needs
Internal Evidence
Clinicians’ suggestions
External Evidence
Literature, reports, etc.
Quality
Governance
Use comparative data to
identify best outcomes
Determine standard
order sets, protocols &
decision support rules
External Evidence
Literature, reports, etc.
Analyze data quality
& process/outcome
variations
Generate the
internal evidence
Clinical Analytics
Other Data Sources
Clinical, Financial, etc.
MTTI
EHR & CDS
Electronic clinical data
Clinicians use standard
protocols & orders
in daily care
Optimal State
Clinical, EHR, EDW &
Analytics Teams
Update EHR protocols &
EDW metrics
Enterprise Clinical Teams
Act on performance
information
Executive & Clinical
Leadership
Set expectations for use
of evidence & standards
Best Evidence
Information that
clinicians trust
Standards

Performance
26

Healthcare Analytics Adoption Model
Level 8
Level 7
Level 6
Level 5
Level 4
Level 3
Level 2
Level 1
Level 0
Personalized Medicine
& Prescriptive Analytics
Clinical Risk Intervention
& Predictive Analytics
Population Health Management
& Suggestive Analytics
Waste & Care Variability Reduction
Automated External Reporting
Automated Internal Reporting
Standardized Vocabulary
& Patient Registries
Fragmented Point Solutions
Tailoring patient care based on population outcomes and
generic data. Fee-for-quality rewards health maintenance.
Organizational processes for intervention are supported
with predictive risk models. Fee-for-quality includes fixed
per capita payment.
Tailoring patient care based on population metrics. Fee-
for-quality includes bundled per case payment.
Reducing variability in care processes. Focusing on
internal optimization and waste reduction.
Efficient, consistent production of reports & adaptability to
changing requirements.
Efficient, consistent production of reports & widespread
availability in the organization.
Relating and organizing the core data content.
Collecting and integrating the core data content.
Inefficient, inconsistent versions of the truth. Cumbersome
internal and external reporting.
© Sanders, Protti, Burton, 2013
27

Progression in the Model
Data content expands
– Adding new sources of data to expand our understanding of care
delivery and the patient
Data timeliness increases
– To support faster decision cycles and lower “Mean Time To
Improvement”
The complexity of data binding and algorithms increases
– From descriptive to prescriptive analytics
– From “What happened?” to “What should we do?”
Data governance and literacy expands
– Advocating greater data access, utilization, and quality
The progressive patterns at each level
28

Six Phases of Data Governance
You need to move through
these phases in no more
than two years
29
3-12 months
1-2 years
2-4 years
– Phase 6: Acquisition of Data
– Phase 5: Utilization of Data
– Phase 4: Quality of Data
– Phase 3: Stewardship of Data
– Phase 2: Access to Data
– Phase 1: Cultural Tone of “Data Driven”
Level 8
Level 1
Personalized Medicine
& Prescriptive Analytics

What Data Are We Governing?
30

Master Data Management
The data that is mastered includes:
– Reference data - the dimensions for analysis
– Analytical rules – supports consistent data binding
Comprises the processes, governance, policies,
standards and tools that consistently define and
manage the critical data of an organization to
provide a single point of reference.”
- Wikipedia
31

Data Binding & Data Governance
“systolic &
diastolic
blood pressure”
Pieces of
meaningless
data
115
60
Binds
data to
Analytics
Software
Programming
Vocabulary
“normal”
Rules
32

Why Is This Binding Concept
Important?
Data Governance needs to look for and facilitate both
33
Knowing when to bind data, and how
tightly, to vocabularies and rules is
CRITICAL to analytic success and agility
Is the rule or vocabulary widely
accepted as true and accurate in
the organization or industry?
Comprehensive
Agreement
Is the rule or vocabulary stable
and rarely change?
Persistent
Agreement

Vocabulary: Where Do We Start?
Charge code
CPT code
Date & Time
DRG code
Drug code
Employee ID
Employer ID
Encounter ID
Gender
ICD diagnosis code
ICD procedure code
Department ID
Facility ID
Lab code
Patient type
Patient/member ID
Payer/carrier ID
Postal code
Provider ID
In today’s environment, about 20 data elements
represent 80-90% of analytic use cases. This will
grow over time, but right now, it’s fairly simple.
Source data
vocabulary Z
(e.g., EMR)
Source data
vocabulary Y
(e.g., Claims)
Source data
vocabulary X
(e.g., Rx)
34

Where Do We Start, Clinically?
We see consistent opportunities, across the industry,
in the following areas:
• CAUTI
• CLABSI
• Pregnancy management,
elective induction
• Discharge medications
adherence for MI/CHF
• Prophylactic pre-surgical
antibiotics
• Materials management,
supply chain
• Glucose management in
the ICU
• Knee and hip replacement
• Gastroenterology patient
management
• Spine surgery patient
management
• Heart failure and ischemic
patient management
35

Start Within Your Scope of Influence
We are still learning how to manage outpatient populations
36

In Conclusion
Practice democratic data governance
– Find the balance between central and decentralized
governance
– Federal vs. States’ rights is a good metaphor
The Triple Aim of Data Governance
– Data Quality, Data Literacy, and Data Exploitation
Analytics gives data governance something to govern
– Start within your current scope of influence and data, then
grow from there
37

Creative Commons Copyright 38
Obtain unbiased, practical, educational advice on
proven analytics solutions that really work in healthcare.
The future of healthcare requires transformative thinking
by committed leadership willing to forge and adopt new
data-driven processes. If you count yourself among this
group, then HAS ’14 is for you.
OBJECTIVE
MOBILE APP
Access to a mobile app
that can be used for
audience response and
participation in real time.
Group-wide and individual
analytic insights will be
shared throughout the
summit, resulting in a more
substantive, engaging
experience while
demonstrating the power
of analytics.

Contact Info and Q&A
dale.sanders@healthcatalyst.com
@drsanders
www.linkedin.com/in/dalersanders/
39

Demystifying Healthcare Data Governance 2014

Recommandé

Recommandé

Contenu connexe

En vedette

En vedette (20)

Dernier

Dernier (18)

Demystifying Healthcare Data Governance 2014

Notes de l'éditeur