Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Adam Wilcox, PhDColumbia University
March 24, 2009
OutlineHistoryClinical Data Repository Clinical Data Warehouse
Clinical Information SystemsStage 1: Early computers calculated data in contextStage 2: Client applications provided access to ancillary dataStage 3: Systems began aggregating data from multiple sourcesStage 4: Data storage provided historical view
And analysisStage 5: Workflow applications formalize processes between clinical roles
Clinical Information System Technology Levels
Level 1: Departmental applicationsLevel 2: Internally‐developed integrated systemsLevel 3: Functional vendor‐based systemsLevel 4: Comprehensive clinical information systems
Clinical Information Systems at Columbia University
Began at Stage 3 Pushing a Level 1 system to Level 2Issues
VocabularyData modelingInterfacesDecision supportData processing
Recipient of first Nicholas Davies Award
EMR environment
Replicate
Distribute
Query ReplicateQuery
Production
Databases
EnterpriseRepository/Data Warehouse
Workgroup
Datamarts
Datamarts,Personal, Mobile,Research, Educ.
Query
Workflow orGoal specific
Information Service LayersInformation Service Layers
Handling
Encoding
Routing
AccessAccess
Monitoring
Other Level 2 SystemsIntermountainVAPartners RegenstriefVanderbilt
Level 3 SystemsCernerEpicEclipsysGEMcKesson
Challenges at ColumbiaMoved from Stage 3 through Stage 4 to Stage 5Purchased a vendor system (Level 3)How to get to Stage 5 and Level 4?
Challenges at CPMC/CUMC/NYPH/WCMC
In 1998, merged two academic medical centers into NewYork Presbyterian Hospital
Columbia Presbyterian campus became Columbia University Medical CenterNew York Hospital became Weill Cornell Medical Center
Currently 4 different electronic health recordsEclipsys (WCMC)Eclipsys (CUMC)Epic (WCMC)Allscripts (CUMC)
Departments (admin)(Clin) My Medical File (Clin) Mobile Telemetry (CUMC) Pyxis (WMC/CUMC) - OR
NYPNYP
ColumbiaColumbia
Weill CornellWeill Cornell Information Service(Adm),Eagle , BIS,(Clin) Eclipsys XA WebCIS, Cisyphus.(Clin) Amicas,GE PACS(Clin) Imnet, Surgical Mgr(Adm ) Lawson, P’soft, etc.(Clin ancil) Cerner, MisysMPI ,Egate , Data Warehouse Network , Desktop, Security
Departments (clin)Climacs (WMC) - Med(Clin) CopathHelix (CUMC) - SurgeryMany others.
ITSNYP EmailNYP CalendarWMC NetworkWMC Email, etc.
Cornell PO(Adm) IDX (Clin) Epic
Sponsored Hospitals
Sponsored Hospitals
Core & ITSNYP/WMC Network
Core CUMC Network
Columbia PO
(Adm) IDX(Clin) Allscripts
ResearchResearch
CUIT CU Email
DepartmentalDepartmental
Local Clinical and Admin Systems
Clin Trial Velos
Pub Relations www.nyp.org Development DB
Heath Plan, etc.
Heath Plan, etc.
LabTestLabTest
RogosinRogosin
Psych InstPsych Inst
Clearing HousesClearing Houses
St Luke’sSt Luke’s HarlemHarlem
HSSHSS
CUBHIS CUMC CalendarDesktop, etc.
Also used by some sponsored hospitalsContractual/Outsourced relationship Dept BillingDept Billing
TSI, EagleTSI, Eagle RegulatoryRegulatory
NYP Computing Environment
Sep 2007
Service CorpService Corp
Service CorpService Corp
Security
NYQNYQ NYMNYM CHOBCHOB Etc.Etc.Presbyterian New York MSCHONY Westchester Allen Pavilion
PresbyterianPresbyterian New YorkNew York MSCHONYMSCHONYWestchesterWestchester Allen Allen PavilionPavilion
Internet
Internet
& CU MS
Eclipsys
(CUMC)
Eclipsys
(WCMC)
Allscripts Epic
Problems with Integrating to Application Databases
Must model each system multiple timesIncreased effort and complexity
Overloading workflow databasesProtecting external data consistency (no updates)Increased complexity of data protectionBringing in data for a new patient
When to pull data inInterfaces don’t naturally pull in historical data
Increases complexity as move toward RHIOs
Eclipsys
(CUMC)
Eclipsys
(WCMC)
Allscripts Epic
Clinical
Data
Repository
Benefits of CDROnly model data from source systems onceCommon data storeData are read only
Optimized for readHistorical data includedWeb‐based viewer adaptable to multiple applicationsAdaptable to future health information exchange effortsPlatform of innovation
Optimized for RetrievalRelational structure can be difficult to query for both data and context
Gathering multiple elements requires multiple table joinsGood for data storageGood for aggregating across multiple patients
Event‐based model good for querying across data types
Data organized according to patientNot good for querying across patients
Retrieval optimizationParadigm shift in how data are used
Paper records mainly for primary useElectronic allows secondary useSecondary use can be multiple times
0%
1 0%
2 0%
3 0%
4 0%
5 0%
6 0%
7 0%
Jan 01 2008
Jan 11 2008
Jan 21 2008
Jan 31 2008
Feb 10 2008
Feb 20 2008
Mar 01 2008
Mar 11 2008
Mar 21 2008
Mar 31 2008
Apr 10 2008
Apr 20 2008
Apr 30 2008
May 10 2008
May 20 2008
May 30 2008
Jun 09 2008
Jun 19 2008
Jun 29 2008
Jul092008
Jul192008
Jul 29 2008
Aug 08 2008
Aug 18 2008
Aug 28 2008
Sep 07 2008
Sep 17 2008
Sep 27 2008
Oct 07 2008
Oct 17 2008
Oct 27 2008
Nov 06 2008
Nov 16 2008
Nov 26 2008
Dec 06 2008
Dec 16 2008
Dec 26 2008
Jan 05 2009
Jan 15 2009
Jan 25 2009
Feb 04 2009
Feb 14 2009
Feb 24 2009
WebC IS tab per XA u se
Us er s
M RNs
Rec s
0%
10%
20%
30%
40%
50%
60%
70%
Jul 01 2007
Jul 17 2007
Aug 02 2007
Aug 18 2007
Sep 03 2007
Sep 19 2007
Oct 05 2007
Oct 21 2007
Nov 06 2007
Nov 22 2007
Dec 08 2007
Dec 24 2007
Jan 09 2008
Jan 25 2008
Feb 10 2008
Feb 26 2008
Mar 13 2008
Mar 29 2008
Apr 14 2008
Apr 30 2008
May162008
Jun 01 2008
Jun 17 2008
Jul 03 2008
Jul 19 2008
Aug 04 2008
Aug 20 2008
Sep 05 2008
Sep 21 2008
Oct 07 2008
Oct 23 2008
Nov 08 2008
Nov 24 2008
Dec 10 2008
Dec 26 2008
Jan 11 2009
Jan 27 2009
Feb 12 2009
Feb 28 2009
0
1000
2000
3000
4000
5000
6000
Eclipsys MRNs
Tab %
1994: Created, sponsored by Columbia University Department of Medical Informatics and Office of Clinical Trials◦
Populated with data from existing clinical data repository◦
Supporting clinical research1998: Columbia + Cornell = NewYork Presbyterian Hospital◦
Warehouse funded by NYPH◦
Goal to incorporate and provide data across whole system2004: Formal analysis of CDW user needs by Clinical Quality and Information Technology Committee (CQIT)◦
Creation of Data Warehousing Subgroup◦
Need to bring together disparate clinical data sources◦
Need to manage user requests for data
CUMC/NYP Clinical Data Warehouse History
Uses of the WarehouseClinical research queriesManagement reportsClinical trial recruitment
CDW Content IssuesBegan as a copy of the repository
Data already gatheredMainly for research queries
Some data marts built for common queriesAbility to query rapidly across patients increases security risk
Goal of Access PolicyProvide broader access to data
Central control is resource limitedAllow collection of more data sources
Reassure data stewards Three separate institutionsData ownership not completely defined for all data
CDW StructureIdentifying data
Patient identifying informationMain data
Event tables for clinical repositoryLookup tables
Vocabulary translationContains no patient data
Specialty data marts
Access PolicyIdentifying data
Most restrictedCreate a research identifier to replace the patient IDAllow access to only ResearchID, sex, birth date (month and year only), marital status, race, death status
Specialty dataAccess policy defined by data steward
Patient clinical dataNo access to text dataModified dates
Lookup tablesFull access (contain no patient data)
Access PolicySpecific patient information
Sometimes needed to create initial queriesAnalysts get access only to a randomly selected subsetAccess request through supervisor
De‐identified patient dataTest patientsFull access given
1994: Created, sponsored by Columbia University Department of Medical Informatics and Office of Clinical Trials◦
Populated with data from existing clinical data repository◦
Supporting clinical research1998: Columbia + Cornell = NewYork Presbyterian Hospital◦
Warehouse funded by NYPH◦
Goal to incorporate and provide data across whole system2004: Formal analysis of CDW user needs by Clinical Quality and Information Technology Committee (CQIT)◦
Creation of Data Warehousing Subgroup◦
Need to bring together disparate clinical data sources◦
Need to manage user requests for data
CUMC/NYP Clinical Data Warehouse History
Analysis of ChallengesData in vendor‐based transactional systemsCould not query across transactional systemsUsers needed help in defining their needsMature initiatives required more robust data solutions
INTEGRA
TION SERV
ICES
SOURCE
Systems
SOURCE
Databases
REPLICATED
Databases
AD HOC
Complex
Analytical
Queries
RECURRING
Clinical Care
Reporting
&
Business
Analysis
&
Research
__
Online
Analytical
Processing
(OLAP)
VIRTUAL CLINICAL DATA WAREHOUSE
DATAMARTS
DM
DM
DM
DM
A
B
C
A
B
C
Reports
OLAP
Research
Goal Task Use User ToolSix
Sigma
Cost/
Instance
Instances Required
Answer a
specific
question
Ad hoc
query
Research Researcher SQL Define + +++++Defined
request
Observe trendsRecurring
query
Management
reports
ManagerReporting
application
Measure ++ ++++Available
owner
Identify
dependencies
OLAPOperational
analysis
AnalystAnalytics /
Data cubes
Analyze +++ +++Content
expert/
analyst
Assist decision
making
Dashboard
display
Point of careClinical
team
Registries Improve ++++ ++ Pilot site
Automate
processes
ApplicationDecision
support
Clinician/
Role
EMR
application
Control +++++ +Institutional
sponsor
INTEGRA
TION SERV
ICES
REPLICATED
Databases
VIRTUAL DATA WAREHOUSE
DATAMARTS
DM
DM
DM
A
B
C
Ad-Hoc Queries–
QuestionsResearch Define
Recurring–
Automated Queries
Management Reports Measure
OLAP–
Analytics
Operational Reports Analyze
Dashboards Point of Care Reporting Improve
ApplicationsDecisionSupport Control
DATA WAREHOUSE TOOLS
Black BeltSix SigmaApproach
ConclusionIntegrating clinical data repository view into workflow applications can improve useAccess policies need to isolate data to reassure data use from different stakeholdersData access tools need to account for users’ evolving data needs along the quality improvement life cycle