The Aggregated Health Information Project: Building October 27, 2006

1
The Aggregated Health Information
ProjectBuilding October 27, 2006
2
Aggregated Health Information Project Background

• Follow-on project to HNData 1999-2004
• Builds on lessons learned during HNData project,
  and from independent reviews
• Key Recommendations
• Move forward with data warehousing under a
  consolidated knowledge development umbrella
• Articulate a vision and strategy and get buy in
• Develop a data culture in the Ministry
• Establishes a broad direction for information
  management at the Ministry

3
Vision

• A provincial health information management
  framework that
• Shifts to patient-centred analysis, sensitive to
  the user, task, responsibility and location
• Moves beyond information to knowledge management
  and development
• Proactively supports health system with new
  information resources and techniques

4
(No Transcript)
5
(No Transcript)
6
Health Information Management Framework
System- Wide Models
Complex Analysis
Service Delivery
Patient
Population
Derived Episodes
Derived Groups
Population
Information Perspectives
Derived Patient Groups
Providers Services
Environment
Organiz- ations and Facilities
Patients
Simple Analysis
Geo-based or Survey based Measures
Finances, Resources, Capacity
Events Associated Measures
Source Data
7
Health System Measurement and Management
Information
Projected and Recorded Demand
Service Demand
Pop. Health Status
Service Delivery
Trends, Patterns, Performance Measurements, Projec
tions
Population- Based Health Status
Distribution, Volume, Cost
Service Outcomes
Event / Case-Based Health Status
8
AHIP Key Objectives

• Enable a sector-wide view of the knowledge and
  information required to understand the whole
  context of health and related service delivery in
  BC
• Bring data, information and knowledge into this
  view on an incremental, priority driven basis and
  integrate all relevant data and information over
  time
• Provide tactical value along with strategic
  deliverables

9
AHIP Key Objectives (contd)

• Concurrently and incrementally extend support for
  data, methodology, and tooling needs of
  information consumers
• Build in all necessary business controls and
  governance processes (e.g. data quality
  assurance access control)
• Use metadata to drive and support the development
  and delivery processes

10
Our Challenges the Ongoing Use of the Data

• Information is typically produced in response to
  an issue rather as an ongoing resource for
  improving corporate knowledge
• Lack of a proactive, comprehensive means to
  undertake the surveillance of trends in health
  status, need for service, efficiency of the
  system and health outcomes

11
Health System Knowledge Model

• Objective To provide a formal, descriptive, and
  extensible outline of the uses of aggregate
  information in supporting the healthcare system,
  in order to
• Contextualize specific business issues, concerns
  and questions
• Promote increasingly well-informed, effective and
  consistent analysis and reporting across the
  scope of information use in the system
• Support system users in identifying, describing,
  finding, retrieving and/or building packages of
  information appropriate to their needs.

12
Uses of Aggregated Health Information
Resources, Finances, Capacity
Service Assessment Events
Evaluation
Population Health
Outcome Analysis
Planning
Health and disease Surveillance
Service Utilization
Performance Measurement
Patient Demographics
Planning
Service Utilization Analysis
Needs Analysis
Population Health Assessment
Program / Policy Evaluation
Service Delivery Profiling
Geo-based or Survey based Measures
Economic Evaluation
13
Knowledge Model Principles

• Focus first on the business context and
  questions, then on furnishing information
• Business questions can be described in terms of
  categories of enquiry that share common
  approaches and methodologies, (e.g. drug
  utilization, lab utilization, physician
  specialist utilization ),
• Business contexts are highly interdependent
• Consistent methodology and a consistent
  (dimensional) approach to data structure can
  provide a great deal of leverage for consistent
  business understanding and analysis.

14
Engaging with Ministry Information Users

• Lead with executive engagement
• Understanding strategic directions
• Communicate vision and potential
• Examine current practices
• Identify new requirements and opportunities
• Business wins and high value products
• Capacity to use information and tools
• Information delivery processes

15
MOH Program area Requirements
Program Executive
Business Questions Priorities Decision Making
Strategic Reporting
Current Assessment
Minimum Reporting
Requirements Business Continuity Communication Sta
ff Participation
Strategic Plans
Standard Methods
Available Data
Working Group
Capability Development
16
Developing the Framework
17
Developing the Framework
What are the real questions? The relevant
measures? The best data sources and algorithms?
18
Developing the Framework
Provide Appropriate Access Tools for Results and
Audience
Support with a formal catalogue of requirements,
questions, measures, algorithms, and information
resources
Provide access to intermediate and Summary
results that are formal, reliable, current
Deliver integrated, standardized, quality-assured
source data
19
The Information Catalogue Metadata
Business Focus Areas
Explanatory and Contextual Notes
Business Questions
Information Products
Measures
Intermediate Resources
Derivation and Aggregation Rules
Standardized Data Structures
Transformation and Quality Rules
Data Sources
20
The Information Catalogue Metadata
Business Focus Areas
Explanatory and Contextual Notes
Knowledge model focus
Business Questions
Information Products
Information delivery focus
Measures
Intermediate Resources
Derivation and Aggregation Rules
Standardized Data Structures
Data acquisition and detailed analysis focus
Transformation and Quality Rules
Data Sources
21
Technology Framework Components
Information Delivery (Research and End Users)
Source Systems
Data Integration Quality Assurance
Ministry, Health Authority Other External
Sources
Staging Area
Reference Space for Dimensions
Detailed Facts Dimensions
Tool Specific Data or Views
Aggregate and Derived Data
Simple
Complex
Metadata Services
Information Resource Catalogue
22
Information Delivery Principles
Information Delivery (Research and End Users)
Source Systems
Data Integration Quality Assurance
Information updates are available in a timely
fashion
Ministry, Health Authority Other External
Sources
Staging Area
Reference Space for Dimensions
Detailed Facts Dimensions
Tool Specific Data or Views
Aggregate and Derived Data
Information provided is consistent
Simple
Information delivery systems are responsive
Complex
Authorized data access is enabled
Information is of appropriate quality
Metadata Services
Metadata is as important as data
Information Resource Catalogue
23
New reporting capabilities with integrated data
and dimensional model

• Layered groups of identifiers and classifiers
• E.g. patient, provider, time periods,
  geography, service types, morbidity / health
  status, age bands
• Used for
• Validating fixed fields in data
• Consistently reporting and linking across sources
• Rolling up / drilling down results
• Filtering levels of detail for protection of
  privacy
• Providing a common frame of reference for
  defining complex measures, cohorts, etc.
• Adds analysis leverage to data and tools
• Changes the way the work is done!

24
Reference Data Dimensional Clusters
25
Support for Data Linkage

• Extend the base dimension e.g. with patient or
  provider demographics, ICD code description
• Co-ordinate views of aggregate info for same
  reference data set e.g. costs by service type
  for same set of patients
• Co-relate event data / reporting across sources
  (e.g. DAD-MSP)
• Build analytic reference data sets by applying
  consistent criteria across multiple data sources
  (e.g. cohorts, episodes, burden of care
  indicators), for use in broader analysis and
  aggregation.

26
Health Information GIS

• The traditional two broad types of GIS
  applications are
• health outcomes and epidemiology applications,
  and
• healthcare delivery applications.
• Most pertinent for us - the interface (overlap)
  between epidemiological and healthcare delivery
  applications
• For the MOHS, geographic information analysis,
  and the evolution of the data and methodological
  insight required for this analysis to be
  effective, are integral parts of our information
  delivery strategies and plans.

27
(No Transcript)
28
AHIP GIS Strategy

• Apply geographic co-ordinates for residence,
  point of service, etc. to all relevant event
  data, based on geo-location of addresses.
• Establish interoperability with the BC
  Governments Land Resource Data Warehouse (LRDW)
  for physical geography, land use and
  infrastructure data.
• Pursue interoperability between dimensional and
  spatial analysis tools and views of data.
• Provide simplified GIS analysis for specific
  purposes such as disease surveillance and
  facility planning.
• Ensure robust controls on data access for
  protection of privacy with respect to exact
  locations.

29
Access to Data

• Security of data
• Access Authorization Control
• Role based access
• Audit logs
• Privacy Protection Identifiable data on a need
  to know basis only
• Data quality assurance and matching
• Medical intervention
• Audit

30
Strategies for managing data linkage for analysis
(near term)

• Separate views and responsibilities for MATCHING
  from views and responsibilities for ANALYSIS.
• Keep audit logs of all access to personally
  identifiable data by individual users.
• Use metadata, dimensional roll-ups, and data base
  views to manage data access and extract,
  balancing need to know birth dates, addresses
  etc., with identification risk (some manual work
  initially).
• Dont use ID encryption if you can easily use a
  non-reproducible surrogate ID.
• Ensure the underlying system security
  infrastructure is robust, in both technical and
  human terms.
• Define and apply formal criteria for masking
  small result set data.

31
Personal information types
Use in ANALYSIS No value, if uniqueness is
preserved No value High value for geographic
analysis High value for demographic
analysis High value Requirement for
detail (e.g. actual date or address) will vary
with context of use.
Use in MATCHING Highest value within same org.
/ sector (scope of id use) High value where
explicit IDs dont exist, or need corroboration
Very low value for person matching in general

• Explicit Business Identifiers (e.g. PHN)
• Common use identifiers (names)
• Contact information (address, telephone)
• Tombstone demographic characteristics (birth
  date, gender)
• Service / assessment information

32
Technology Framework Components
Information Delivery (Research and End Users)
Source Systems
Data Integration Quality Assurance
Ministry, Health Authority Other External
Sources
Staging Area
Reference Space for Dimensions
Detailed Facts Dimensions
Tool Specific Data or Views
Aggregate and Derived Data
Simple
Complex
Matching
Analysis
Metadata Services
Information Resource Catalogue
33
Strategies for managing data linkage for analysis
(longer term)

• Develop flexible utilities and capabilities for
  delivering fast, iterative, aggregate views of
  data (including statistical and geographic
  aggregates) accurately on demand from detailed
  data, without revealing the detailed data to the
  analyst.
• In extracting detailed, de-identified data for
  external analysis, encrypt surrogate identifiers
  differently for each authorized purpose.
• Establish and apply a code of ethics for data
  access request reviews automate to the extent
  possible.

34
Data linkage across organizations (primarily
government and healthcare)

• Shared Client
• Access to health services and health outcomes are
  highly influenced by other government services -
  community, social, educational, and economic
  factors and
• These services in turn are impacted by the health
  status of their clients and workers
• Data Privacy, Security issues
• personal information highly sensitive
• mechanisms, policies and legislation in
  development

35
How can broader linkage work?

• Assumptions
• Each Ministry (or other organization) maintains
  responsibility for their own data, in terms of
  collection, stewardship and governance
• In sharing among / linking across organizations,
  ensure the same minimum set of info privacy
  controls and policies exist for analytic access
  to linked and unlinked data, in all participating
  organizations.
• The Trust Box
• A secured auditable environment to allow the
  linkage of anonymized aggregated data for
  decision support.
• Provides enhanced functions for segments of the
  data reference / integration layer that are
  common across multiple organizations, and highly
  sensitive
• This would require data from multiple sources to
  be linkable based on common reference data and
  common data models.
• Technically feasible
• Requires legislative and policy enablement
• Enables a virtual warehouse of warehouses

36
AHIP principles underlying the data access
approach

• Explicit, effective linkage with explicit,
  effective governance will support better analysis
  and more complete and auditable privacy than are
  possible with restrictive, ad hoc linkage.
• Data linkage, data quality, and the practical
  basis for dimensional analysis are all closely
  related, and should be managed together.
• Metadata makes this entirely practical.

37
Over-All Strategy Summary

• Consistent management and technical framework
• Apply to all Ministry data analysis and
  information delivery
• Open to integration and analysis across
  organizational boundaries
• Grow the scope incrementally, based on a
  strategic view of data requirements
• Focus first on patient- and provider-based
  information
• Initially with smaller data sources (e.g. DAD)
• Make key cross-data source analysis a priority
• Import useful derived information (e.g. Chronic
  Disease cohorts) even if intending to generate
  them in the longer term
• Focus on strategic data analysis and new business
  value, supporting operational reporting / data
  analysis in the same framework as much as is
  practical

38
Emerging opportunities

• Enhanced ad hoc analysis of detailed data using
  new tools such as Oracle Discoverer
• Packaging and rapid delivery of aggregate
  measures (similar to PURRFECT, web-based)
• Support for more complex analysis using
  anonymized linked data, cohort formation, episode
  analysis, geographic data integration and
  mapping,

39
Thank you Questions or comments are welcome
or forward to the KID/AHIP teamterry.tuk_at_gov.bc.
caian.caesar_at_gov.bc.ca kelly.barnard_at_gov.bc.cac
harles.douglas_at_gov.bc.ca