Health Record Architecture(s) in Germany
Motivation and Architecture Overview
Dr. Jörg Caumanns, Fraunhofer-Institut FOKUS Berlin
Agenda
The „Philosophy“ of the German eCR+PHR paradigm The Patient in Control: The German PHR platformThe EHR/PHR Success Story ??
Google.health -> leaving the German/European market Microsoft HealthVault -> almost invisible in Germany ICW LifeSensor -> suspended CGM Life (Vita-X) -> re-launch, low market visibility Siemens Assignio -> to be suspended? regional EHR networks -> dependend on public funding, few in real operation regional EHR networks -> dependend on public funding, few in real operation NHS in England -> suspended Dutch EHR system -> suspended Czech EHR system -> suspended German telematics infrastructure -> 7+ years late, reduced functionality Many other governmental EHR/PHR activities delayed due to privacy/security issues,The Balance is the Challenge
Privacy
Security
Usability
Flow Integration
Functionality
Value of Benefit
The „One fits All“ PHR/EHR Compromise
Physicians actively accessing patients‘ data (e.g. query-retrieve-pattern)– Who is in control?
– Who is responsible?
Privacy enforcement by Access Control– Patient as the „bad guy“ who
– Patient as the „bad guy“ who
imposes constraints
– Access Management vs. Usability
Static Configuration (Permissions, Identities, Roles, …)– On-Demand Use?
Immediate access -> database Data processing -> database (OLTP) Async. Communication -> eMail Bilateral Data Exchange -> eMail
Long term storage -> archive
Data analysis -> data warehouse (OLAP) Sync. Communication -> phone, chat Multilateral Data Sharing -> FTP, etc.
Different Requirements <=> Different Solutions
Bilateral Data Exchange -> eMail Machine Processing -> XML Publishing Data -> Web Site, Wiki, …
Multilateral Data Sharing -> FTP, etc. Display to Human -> PDF
Cooperation -> Project Place etc.
Why do eHealth architects always asume that a single
health record architecture could serve all needs?
Privacy
Patient in Control
Closed User Group
Consent
Functionality
Spontanous Need
Active Treatment
Usability
Interaction
Data Integration
Privacy
Patient in Control
Closed User Group
Consent
Functionality
Spontanous Need
Active Treatment
Usability
Interaction
Data Integration
No Compromises
Personal Record acc. § 291a SGB V– Citizen interacting with her physicians
– Citizen owns data and has full power
of control
– MoH R&D Project as incubator for a future national solution
Patient in Control
Spontanous Need Interaction
Closed User Group
Active Treatment Data Integration
Electronic Case Record (ECR/EFA)– Members of the care team sharing data
related to a single medical case
– Patient consents to the care team
– Virtual integration of physicians‘ data within regional healthcare networks
Personal Health Record acc.
§
291a SGB V
Asynchronous Interaction Patterns
– Physician->Citizen: Could you please provide me with your medication plan?
– Citizen->Physician: Here you are.
– Citizen->Physician: Could you please provide me with last weeks lab results?
– Physician->Citizen: I’ve updated your nutrition plan. Here it is.
– Citizen->Physician: Here are last weeks blood glucose values.
Modular Patterns:
• Physican requests Data from Citizen
• Phyician provides Data to Citizen
• Citizen requests Data from Physician
• Citizen provides Data to Physician
Synchronous Interaction Patterns
– Physician->Citizen: I would like to register this new medication with your medication plan.
– Citizen->Physician: Could you
please just register this X-Ray with my PHR?
– Citizen->Physician: Yes, that test has already been done. The report is available with my PHR.
– Physician->Citizen: The test shows that you have a cat allergy. Would you like me to set up an allergy passport within your PHR?
Modular Patterns:
• Physican requests data (together with
a patient token); PHR provides data
• Physican provides data (together with
Integration Platform
Vaccination Record …
Medication Plan
Semantic Interoperability Framework Interaction
Online Database …
USB Device
Security Modules (Authorization, …)
Communication and Storage Abstraction Layer
Adapter Adapter Adapter
Interaction Patterns
Semantic Signifiers
The PHR is just a semantics-agnostic platform! Medical use cases are served byapplications (e.g. medication plan, allergy passport, patient diary, ….) which are implemented on top of this platform.
Application semantics are encapsulated as “semantic signifiers”.– Concept map of the application
– Information models (incl. schemas and schematrons) of the application objects
– Information models (incl. schemas and schematrons) of the application objects
– Behavior wrt. create, read, update of application objects
E.g. providing a new medication plan makes this the current medicationplan. Requests are always on the current medication plan only.
Each interaction is bound to a specific semantic signifier! Two special semantic signifiers:Electronic Case Records (ECR) / elektronische Fallakte (EFA)
Essentials
ECR is an initiative which is driven and sponsored by German hospitals The objective is to provide a data sharing plattform for regional care networks– Hospitals as data providers and service providers
– Self-Organization (no central services but PKI)
– Conformant to German Privacy Regulations
– Adaptable to disease/contract-specific care scenarios
– Adaptable to disease/contract-specific care scenarios
Status– Specifications have been published in 2008
– 10+ pilot projects in different regions
– 2 fully operational networks (Munich, Aachen)
– Product support from industry (Siemens, iSoft, ISPro, CrossCheck)
ECR Logical Model
Scenario: Someone is ill and subject to co-operative treatment
–
The kinds of relevant data and the scope of the treatment can be defined
purpose specific data collection
–
co-operating physicians and organizations need access to all data which is
relevant for the patient’s treatment
patient gives consent to this (NO fine grained permissions needed!)
An ECR medical case corresponds to a co-operative treatment
Each medical case is made up from folders and administrative cases (which are
folders, too)
An ECR administrative case corresponds to an inpatient visit that is part of the
treatment -> automated linkage of visits and ECR folders
Medical data is registered within folders
ECR as a Federated System
A u the nt Auth entis ierun g Authe ntis ieru ng e n tis ie ru n g A u to ri s ie ru n gECR Security Architecture (Overview)
Identity Provider ECR Admission Token Service ECR Record Registry Authenticate Users / Assign Attributes Create Admission Codes Discover Patient’s Records Enforce Access Policies ECR Document Registry ECR Access Token Service Assign Access Policies ECR Document Registry ECR Policy Token Service Obtain Access Policies Id e n ti R e ID lic y PatientID RecordID Identity Assertion Admission Assertion Access Assertion ECR Document Repository IDM Policy Assertion tit y / A ttr ib u te s Admission Lists R ec ord Id en tifier Acc ess Pol icy ID Policy Registry A c c e s s P o liAdmission Code RecordID PolicyID PolicyID Access Policy UserID Attribute
RecordID Metadata SecretHash
ECR Assertion Chaining
Identity Provider
Admission Token Service
produces produces co n su m e s su m e s identit y assertion admission assertion access assertion efe re n ce re fe re n ce patient Access Token Service
Policy Service produces produces co n s co n su m e s su m e s access assertion
policy assert ion refe
re n ce re consent w ho? w hat?
ECR Building Blocks
Hospital acting as ECR Provider
ECR Consumer
„ECR in a Box“ Deployment: Full
encapsulation of ECR Security
Record Management: Logical Operations
Register Consent Deprecate Consent Register Episode/Visit De-Register Episode Update Episode Metadata …. Register Document De-Register Document Update Document Metadata– incl. diagnosis-specific Metadata
…Dedicated service interface per management operation?
ECR R-MIM
The full ECR Information Modell can be implemented as an CDA compliant R-MIM New ECR instances can be set up by sending a respective CDA to the ECR Box– E.g. using a standard doctor’s letter “Dear ECR Box….”
The current status (content) of an ECR instance can be exported as a set of interlinked CDA documentsCDA documents
A ECR instance can be managed by adding/deleting/updating sections and entrieswithin the respective CDA document
– E.g. updating the purpose is done by updating the respective entry within the CDA’s diagnosis section
Diagnosis specific extensions can be implemented and managed by simply addingECR R-MIM
Header
Patient
Initiator
eCR Provider
eCR Object
Contract Docs.Roles
Care Providers
Body
Contract Docs. Purpose HCP Visits DiagnosiseCR-V
eCR-specific sections (optional) CAVE Docs.ECR Visit (Folder) R-MIM
eCR-V
Header
Patient
Care Giver
eCR Object
DiagnosisInpatient Visit
eCR-V
Body
Admission Diagnosis Medical Documents Diagnosis eCR-specific sections (optional) Contract Docs.Nesting of Standards
IHE PCC PHR Update
ECR R-MIM (Subset of CDA R-MIM)
P
ro
vi
d
e
r
In
te
rf
a
ce
C
D
A
P
ro
ce
sso
r
E
C
R
S
e
rvi
ce
s
OMG/HL7 RLUS
SOAP
HTTPS
Example: Registration of a Document with an ECR
<ecr:InitializeRLUSECRBoxRequest> <!-- eCR Document --> <cda:ClinicalDocument> ... </cda:ClinicalDocument> <RLUStypes:RLUSInitializeRequestSrcStruct> <RLUStypes:RLUSsemantic-signifierName> ecr-document </RLUStypes:RLUSsemantic-signifierName> <RLUStypes:SecurityContext>...</RLUStypes:SecurityContext> <RLUStypes:SecurityContext>...</RLUStypes:SecurityContext> <RLUStypes:InitializeContext> <!—visit identifier --> <cda:encompassingEncounter><cda:code code="INP„ codeSystem="1.2.276.0.76.3.1.81.81.5.4" /> <cda:id root="..." extension="..."/>
<cda:effectiveTime NullFlavor="UNK" /> </cda:encompassingEncounter>
</RLUStypes:InitializeContext>
ECR Box (Plug & Play Deployment)
Crosscheck Networks
Frans Halslaan nr 9
5143 GJ Waalwijk
Conclusion
EHR/PHR is just a platform– Multiple, simple and easy-to-use platforms vs. making compromises on usability, functionality and privacy
– Access control means must be integrated with the logical architecture
The 1st challenge is on extensibility and adaptability– Disease specific care networks, Quality management, reporting procedures, …
– Disease specific care networks, Quality management, reporting procedures, …
– Semantics-aware domain/application models as basis for access operations (e.g.
query for current medication)
– Application-specific adaptation of CRUD behavior
The 2nd challenge is on EHR/PHR administration and operation– The deeper the EHR is integrated with hospital IT the more management