Creating a Successful Clinical Reporting Tool

1
Creating a Successful Clinical Reporting Tool
Executive Information Decision Support
(EIDS) United States Department of Defense
Military Health System
2
Presenter

• Lieutenant Colonel Bob Aarhus
• Program Manager
• Executive Information Decision Support (EIDS)
• TRICARE Management Activity
• United States Military Health System

3
Objectives

• List common challenges that can be experienced by
  a clinical reporting tool as its source
  transactional system matures
• Identify tasks that should be undertaken by
  organizations to promote success of a clinical
  reporting tool

4
Some Terms

• Transactional System
• Operational system for an enterprise
• Day-to-day, usually one record at a time
• Optimized for response
• Data Warehouse
• Collection point for data from multiple
  operational systems
• Subject-oriented, integrated, time-variant,
  non-volatile data store
• Few direct users
• Data Mart
• Provides views of data from a data warehouse,
  usually in the context of a single business
  process
• Limited scope (normally)
• Multiple users

5
Key Components

• Composite Health Care System (CHCS)
• transactional system adjunct to paper record
• AHLTA
• longitudinal electronic health record
• Clinical Data Repository (CDR)
• stores transactional data from AHLTA
• Clinical Data Mart (CDM)
• receives CDR data originally Clinical Data
  Warehouse

6
United States Military Health System

• 9 million beneficiaries worldwide
• Active duty, reserves, families, retirees,
• 70 hospitals, 400 clinics

7
Developing Applications
Military Services Provide user requirements
Joint Capabilities Integration and Development
System Identify gaps
TRICARE Establish policy and guidance
Office of the Chief Information Officer Manage
the portfolio
Integrated Requirements Design Decompose
requirements
Program Offices Deliver the Applications!
Department of Defense Establish
acquisition framework
U.S. Congress Enact law (Clinger- Cohen),
oversight
8
AHLTA The Cornerstone

• Militarys longitudinal electronic health record
• Beneficiary health data available on demand
• Replacement for paper health record, adjunct
  electronic health information system
• Structured notes core of the physician interface

9
First Challenge Query the Clinical Data
Repository (2002)

• Reporting an important element in AHLTA
• Result desired retrieve data from the Clinical
  Data Repository and format it for reporting
• Outcome Clinical Data Repository unstable
  queries caused reduced transactional performance

Lesson Learned Transactional systems are not
optimized for most reporting functions
10
Is Clinical Reporting a Priority? (2003)

• Clinical Data Warehouse conceived as technical
  demonstration to
• Extract data from the Clinical Data Repository
• Report that data through a Business Objects
  interface
• but cost/schedule were unacceptable
• Performance is not an issue
• Low-end hardware donated for proof of concept

Lesson Learned Evaluate costs and benefits at
all levels of the enterprise before program
initiation
11
The Basic Concept
AHLTA CDR (Back Up)

• Near real time report of unit readiness and
  population data
• Medical encounter data from the Clinical Data
  Repository in a patient-centric format

DB Server
ETL Routines
Business Objects Server
AHLTA User
MTF Security Server (Snareworks)
12
Moving Ahead

• Renamed Clinical Data Mart
• Third normal form (3NF) relational Oracle
  database
• Interface supports ad hoc queries built
  primarily for business analysts
• First code release run against the Clinical
  Data Repository
• Clinical Data Repository stability still an issue
• Stop Work ordered until issues resolved

Lesson Learned Transactional systems should be
fully deployed and functional first
13
The Historical Extract

• Legacy data was pulled from the Clinical Data
  Repository to populate the Clinical Data Mart
• 24 month extract/ingest cycle at about a 31
  ratio
• 3 months of data pulled in 1 month real time
• Critical Strategic Change Promote AHLTA

Lesson Learned Appropriate funding is vital.
Priorities will change be flexible.
14
Too Slow

• Users reaction too slow
• Minutes to hours to return standard queries
• Excessive table joins, up to two dozen per query
• Aging hardware (4 CPU N-Class Server)
• 2006 First year of funding
• New Clinical Data Mart environment procured
• Business Objects servers
• IBM high-end processors

Lesson Learned Data marts and 3NF Model death
spiral there are performance metrics even
analysts will not tolerate
15
AHLTA-CDM Architecture
CDM Ingest Server
CDM DB Server
AHLTA CDR
CDM ETL Server
Extract changes (PL SQL)
Flat files NFS disk
Ingest data Informatica
CDM Data Oracle DB
CDR CDR HDD Medcin Oracle DB
Flat files
Reporting DB Copy A
Reporting DB Copy B
Build Indexes
Old Hardware
New Hardware
16
Brightening Future

• New equipment significantly improves performance
  
• Data model still inefficient
• Data quality issues identified
• Growing population of enthusiastic users
• Portion of 2007 funds restored
• Will support Wounded Warrior reporting
  initiatives
• As configured, not ideal, but an adequate start
  for future development

Lesson Learned With clinical reporting, the
application may be problematic but the capability
is priceless
17
AHLTAs Clinical Data Mart
18
Appointment Count by Pain Scale
19
Female Patient Totals by Body Mass Index Category

20
Encounter Status
21
What is Success?
22
Did We Succeed?

• Not yet, but
• Extraction capability is present and can be
  expanded
• Data model revisions ongoing
• Good data quality evaluation tool for AHLTA, a
  mission-critical application
• Clear justification for funding a Clinical Data
  Warehouse or Clinical Data Warehouse-like
  structure

23
Lessons Learned

• Transactional systems are not optimized for most
  reporting functions
• Evaluate costs and benefits at all levels of
  enterprise before program initiation
• Operational systems need to be operational first
• Appropriate funding is vital
• Priorities will change be flexible
• Data Marts 3rd Normal Form Model death spiral
• There are performance metrics even analysts will
  not tolerate
• With clinical reporting, the application may be
  problematic but the capability is priceless

24
Questions and Comments

• Visit HIMSS Exhibit Booth 701
• EIDS Web portal
• www.eids.ha.osd.mil
• MHS Help Desk
• 1-800-600-9332
• eids_at_mhs-helpdesk.com