View
289
Download
0
Category
Tags:
Preview:
Citation preview
Proprietary and Confidential © 2015 Health Catalystwww.healthcatalyst.com
July 22, 2015
Powering Medical Research With Data: The Research Analytics Adoption Model
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential 2
Why are we here?
• Sick and healthy patients alike want better care. Research helps determine what we mean by better care.
• There is incredible waste in research. We can greatly reduce this waste with data and analytics.
• The future: precision medicine aims to deliver the right care to right patient at the right time
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential3
To get to Precision Medicine• Improve research – identify underlying molecular causes of
disease and refine treatments
• Improve care delivery – deliver to care guidelines and adapt
• Create better coordination between care delivery and research
3
Time
Measured in Weeks or Months
Habit of allFront-line Clinicians at
Every FacilityNew Clinical or Operational Best Practice Knowledge Discovered
Measured in Years
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential 4
Agenda
• Review Research Process
• Review Roadblocks to Efficient Research
• Present Research Analytics Adoption Model
• Conclusion
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential 5
Agenda
• Review Research Process
• Review Roadblocks to Efficient Research
• Present Research Analytics Adoption Model
• Conclusion
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential 6
Research ProcessHypothesis Generation
• Previous research• Current literature• Exploratory analysis
Cohort Exploration
• How many patients match my criteria?
• Do I have enough patients to do my study?
Grant Application
• Data from cohort exploration should be used
• Historical recruitment data should be used if positive
IRB Application
• Regulatory requirement• Identify population, protocol, and
data needs
Patient Recruitment
• Who are my patients?• Where can I find them?• How will I collect their consent
Data Collection• Prospective data collection
(questionnaire)• Retrospective data collection
(data pull)• Approved by healthcare
organization
Data Analysis• Statistics• Advanced tools for
genomics• Needs to be secure
Publication• Conclusions are compiled• Manuscript submitted• Nobel Prize received
Translate to Clinical Practice
• How can this discovery now be used to treat patients?
• Work closely with care delivery system
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential 7
Hypothesis Generation
Previous research
Current literature
Exploratory analysis
Cohort Exploration
How many patients match my criteria?
Do I have enough patients to do my study?
Grant Application
Data from cohort exploration should be used
IRB Application
Regulatory requirement
Identify population, planned interventions, and data
Patient Recruitment
Who are my patients?
Where can I find them?
How will I collect their consent
Data CollectionProspective data collection
(questionnaire)
Retrospective data collection (data pull)
Approved by healthcare organization
Data Analysis
Statistics
Advanced tools for genomics
Needs to be secure
Publication
Conclusions are compiled
Manuscript submitted
Nobel Prize received
Translate to Clinical Practice
How can this discovery now be used to treat patients?
Work closely with care delivery system
Research Process – Roadblocks (Waste)
Technical Insufficient Exploratory
Tools
OrganizationalNo process for release of data
Technical Lack of single source for dataInsufficient self-service tools
Insufficient tools to support data release process
OrganizationalInstitutional restrictions
TechnicalInsufficient data and tools to
find patients
OrganizationalSlow IRB
TechnicalInsufficient Exploratory Tools
Insufficient tools to support IRB process
OrganizationalLack of ‘deployment
system’Research not aligned with
care improvement initiatives
Technical Insufficient Exploratory
Tools
TechnicalInsufficient Exploratory
Tools
TechnicalInsufficient analysis
tools/platform
OrganizationalInsufficient skillset
OrganizationalLack of support for
manuscript preparation.
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential 9
Research Process - Other Considerations
Data Integration
Operational ReportingProspective data collection
(questionnaire)
Retrospective data collection (data pull)
Approved by healthcare organization
Technical Too much time spent cobbling operational
reports. Takes away from research
Sharing DataProspective data collection
(questionnaire)
Retrospective data collection (data pull)
Approved by healthcare organization
Technical Data coordinating centers often lack infrastructure
for efficient sharing across sites
Data IntegrationProspective data collection
(questionnaire)
Retrospective data collection (data pull)
Approved by healthcare organization
Technical Experimental data siloed
and limited ability to combine with clinical
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential 10
Agenda
• Review Research Process
• Review Roadblocks to Efficient Research
• Present Analytics Adoption Model
• Conclusion
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential11
Level 0 Manual dataset generation
• Data is delivered through operational analysts
• Research data requests typically prioritized very low
• Often no set research process or infrastructure
• Result: frustrated analysts and frustrated researchers
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential12
Subject Area Data Marts
Linking & StandardizationCommon Linkable Identifiers, Patients, Labs, Encounters, Diagnoses, Medications, etc.
ContentPopulation Definitions (800+), Hierarchies, Comorbidities, Risk Stratification, Attribution
Source Marts
EMR
EMR Financial IDEA BioBank Clinical TrialsResearch Registries
Enabler for all Levels > 0: Data Warehouse
Financial IDEA BioBank Clinical TrialsResearch Registries
e.g. Epic, CernerNextGen
e.g. EPSi, Peoplesoft,
Lawson12
OperationsResearchQuality
(Custom data collection tool)
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential13
Level 1 De-identified tools and data marts
• De-identified applications minimize data access roadblocks
• Used for preparatory to research activities
• Can provide a starting point for data analysis
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential14
Level 2 Delivery of customized data sets
Data CollectionProspective data collection
(questionnaire)
Retrospective data collection (data pull)
Approved by healthcare organization
Technical Siloed data
Data CollectionProspective data collection
(questionnaire)
Retrospective data collection (data pull)
Approved by healthcare organization
Technical Insufficient tools to
support data release process
Data CollectionProspective data collection
(questionnaire)
Retrospective data collection (data pull)
Approved by healthcare organization
Technical Insufficient self-service
tools
Data CollectionProspective data collection
(questionnaire)
Retrospective data collection (data pull)
Approved by healthcare organization
Organizational No process for release of
data
• Clear guidelines established for data release• Appropriate data stewards involved• Dedicated research analysts who understand
institutional data, regulatory rules, institutional rules
• Datasets delivered in agile, consultative manner• Tools to facilitate data request workflow
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential15
Level 3 Study recruitment facilitated by EDW
• Provide tools that allow an investigator to define a population of eligible patients
• Improve death status using external data, if necessary• Provide up-to-date scheduling information to recruiters
via mobile device• Provide option for electronic consent• Capture ‘do not contact’ requests• Coordinate patient approaches to prevent ‘study fatigue’
Patient RecruitmentProspective data collection
(questionnaire)
Retrospective data collection (data pull)
Approved by healthcare organization
Technical Insufficient data and tools
to find patients
Patient RecruitmentProspective data collection
(questionnaire)
Retrospective data collection (data pull)
Approved by healthcare organization
Organizational Institutional restrictions
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential16
Level 4 Research-specific data collection is centralized
• Move away from Excel and Access as data collection tools
• Provide infrastructure to support self-service data collection form creation/administration
• Data collection tool should pull from and push to EDW• Pull master data• Push collected data into EDW
• Apply appropriate security to the tool as well as EDW extract
Data CollectionProspective data collection
(questionnaire)
Retrospective data collection (data pull)
Approved by healthcare organization
Data Analysis
Statistics
Advanced tools for genomics
Needs to be secure
TechnicalSiloed data
Technical Insufficient analysis
tools/platform
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential17
Level 5 Automated reporting of research operations
• Pull key research systems into EDW• Clinical trials management system• Patient recruiting system• Electronic IRB system
• Provide key metrics• How many active studies do we have?• How many of our patients are enrolled in trials?• How many of our patients do not want to be involved in
research?
Operational ReportingProspective data collection
(questionnaire)
Retrospective data collection (data pull)
Approved by healthcare organization
Technical Too much time spent cobbling operational
reports. Takes away from research
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential18
Level 6 Biobank/Genomic data integration
• Leverage incredible richness of clinical data set with biological inventories and data• Easily answer question: How many bio samples do I have for
female COPD patients over the age of 65 across all repositories?
• Provide a platform for genomic discovery• Easily answer question: What populations are enriched for a
given set of gene variants?• Easily answer question: What gene variants are enriched for a
given population?• Be aware of regulations on genomic data
Data IntegrationProspective data collection
(questionnaire)
Retrospective data collection (data pull)
Approved by healthcare organization
Technical Experimental data siloed
and limited ability to combine with clinical
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential19
Level 7 Multi-site data sharing
• Provide as much automated extracts to national registries as possible
• Coordinating centers leverage data warehousing techniques for data collection
• Coordinating centers have automated intake or federation of data from participating sites
• Coordinating centers provide exploratory tools to analyze combined data set
Sharing DataProspective data collection
(questionnaire)
Retrospective data collection (data pull)
Approved by healthcare organization
Technical Data coordinating centers often lack infrastructure
for efficient sharing across sites
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential20
Level 8 Translational Analytics
Translate to Clinical Practice
How can this discovery now be used to treat patients?
Work closely with care delivery system
OrganizationalLack of ‘deployment
system’
Translate to Clinical Practice
How can this discovery now be used to treat patients?
Work closely with care delivery system
OrganizationalResearch not aligned with
care improvement initiatives
• Care delivery using analytic tools for population health/care management
• Research aligned with care delivery priorities• Research leverages cohorts, outcome measures from
care improvement while adding in new data• Discoveries from research quickly turned back to care
improvement analytics for deployment across the system
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential21
Translational Research Registries: Discovery
Heart Failure Data Mart
Heart Failure SAM(De-Identified)
GenomicsData Mart
De-identificationToolkit
• Cohort• Outcomes• Process Metrics• Focus: Care Improvement
• Integrated Genomics Data
• Cohort• Outcomes• Process Metrics• Experimental Data• Focus: Discovery
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential22
Translational Research Registries
Heart Failure Data Mart
Heart Failure Study(Identified)
RecruitmentToolkit
• Cohort• Outcomes• Process Metrics• Focus: Care Improvement
• Cohort• Outcomes• Experimental data• Includes genomic data prospectively collected for study• Prospectively collected questionnaire data• Focus: Translational
with IRB
approval
Questionnaire Genomics
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential23
Translational Research Registries: Back to Clinical Practice
Heart Failure Data Mart
Heart Failure Study(Identified)
RecruitmentToolkit
• Cohort• Outcomes• Process Metrics• Focus: Care Improvement
• Cohort• Outcomes• Experimental data• Includes genomic data prospectively collected for study• Prospectively collected questionnaire data• Focus: Translational
with IRB
approval
Questionnaire Genomics
Include new evidence into Care Improvement focused data mart
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential24
Research Analytics Adoption ModelLevel 8 Translational Analytics
Learning health system routinely transitions research data into care delivery guidelines. Variation in patient care is measured, but deliberate and personalized.
Level 7 Multi-site data sharingInfrastructure to participate in multi-site studies and research registries. Coordinating centers use automated tools to collect and combine data from different sites.
Level 6 Biobank/Genomic data integrationBiobank and experimental genomic data are integrated with the EDW. These repositories are no longer limited by data collected at time of sample acquisition.
Level 5 Automated reporting of research operationsSystem-wide research metrics regarding finances and accrual are easily generated and used by researchers. Analytics supporting research billing compliance are provided to the health system.
Level 4 Research-specific data collection is centralized
Data collection tools designed for research can feed directly into the data warehouse. Tools can be pre-populated with approved data from the data warehouse.
Level 3 Study recruitment facilitated by EDWCohort criteria, scheduling data, treating physician data are combined to facilitate study recruitment. Recruitment contact lists are centrally managed to prevent study fatigue.
Level 2 Delivery of customized data sets including clinical notes
Dedicated research analysts deliver custom data sets to researcher. NLP analysis of notes. IRB is engaged in designing data access workflow. IRB templates are designed specifically for data set definition. Workflow tools support approval process.
Level 1 De-identified tools and data martsDe-identified tools allow cohort counting/exploration through user interfaces. De-identified views into the data warehouse provide platform for deeper discovery and hypothesis generation
Level 0 Manual dataset generation Operational analysts carve out small fraction of time for research dataset generation. Still a lot of spreadsheet-based manual chart abstraction.
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential25
Research Analytics Strategy(a starting point)
Time
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential 26
Data Warehouse Considerations for Research
• Critical to provide both de-identified and identified access paths
• Warehouse must provide security model that allow this kind of access
• Organization needs to devise policies that enable access for appropriate users
• Ability to ingest wide variety of research data must be quick and easy
© 2015 Health Catalystwww.healthcatalyst.comProprietary and Confidential 27
Conclusion
• Research-Clinical alignment is a strategic initiative, requires executive engagement and support
• Use the model to evaluate current capabilities
• Turn the model on its side for a starter strategy
• Please email me with feedback! eric.just@healthcatalyst.com
• I mean it!
• THANK YOU for your time today!
Healthcare Analytics Summit 15
Here’s a sneak preview …Industry-leading Speakers
Jim CollinsBest-selling author of Good to Great, Great by Choice, Built to Last, and How the Mighty Fall
Ed CatmullCo-founder of PixarPresident of Pixar and Walt Disney Animation Studios
Daryl MoreyHouston RocketsGeneral Manager and Managing Director of Basketball Operations
Amir RubinStanford Health CarePresident and CEO
Timothy G. Ferris, MD, MPHPartners HealthCareSenior Vice President ofPopulation Health Management
Timothy Sielaff, MD, PhD, FACSAllina HealthChief Medical Officer
Summit highlights3-day AgendaWe’ve increased the time of this year’s summit to allow for more sessions, topics, and networking.
CME Accreditation for CliniciansThis activity has been approved for AMA PRA Category 1 Credits™.
More Case Study SessionsHealth system case studies addressing even more clinical, technical, operational, and financial examples.
Hands-On Experiences Examples, vignettes, and audience-based activities demonstrate principles in fun and memorable ways.
Analytics-Driven EngagementReal-time polling, networking, Q&A, and gamification experiences; plus,i-beacon location technology.
NetworkingExperience networking options that use analytics creatively to help you find and connect with others.
Pre-Summit Classes and TrainingAn early half-day of pre-session classes and training options specifically for Health Catalyst clients.
3X the sessions8 keynotes, 25 breakouts, 25-40 analytics walkabout mini-sessionsf
Early Registration Pricing, Optimized For Teams
Buy 1(save $300)
$395/Pass(through May 31)
Buy 3(save $1,098)
$329/Pass(through May 31)
Buy 5(save $2,000)
$295/Pass(through May 31)
Recommended