Interoperability and Policy Management

James Williams – Ontario Telemedicine Network
Objectives:
Review policy constraints for EHR systems.
2. Traditional approaches to policies in EHRs.
3. CHI consent management architecture.
4. Current research.
1.
Focus:
 Policies pertaining to personal health information.
 Policies may touch upon:
 Consent directives.
 Acceptable uses.
 Permissible disclosure.
 Appropriate safeguards.
 Emergency overrides.
 Retention.
Sources of Policy:
1.
2.
3.
4.
5.
6.
Statutes and regulations
Case law
Codes of conduct
Corporate bylaws
Professional guidelines / best practices
First Nations Sovereignty
Statutes: Privacy
 The most important legislative instruments are the
various privacy and health information statutes.

 Privacy legislation in Canada is based on a set of fair
information practices:
1) Accountability
6) Accuracy
2) Identifying purposes
7) Safeguards
3) Consent
8) Openness
4) Limiting collection
9) Individual access
5) Limiting use, disclosure, retention.
10) Challenging compliance
Statutes:
 Establish a basic rule, and then add exceptions.
 For example, express consent is generally required in
order to disclose information to a third party. But:
 Emergency situations.
 Law enforcement.
 Public health.
 Eligibility for benefits.
 Risk to third party.
Statutes: Private sector privacy laws
Statutes: Health information laws
Statutes: additional laws
 Federal:
 Statistics Act.
 Quarantine Act.
 Provincial:
 Child Protection Act.
 Communicable Disease Act.
 Health Act.
 Worker’s Compensation Act.
 Mental Health Act.
Other sources
 Case Law:
 Eg: Patient has right of access to their own health record.
(McInerney v MacDonald).
 Codes of Conduct:
 Eg: Canadian Medical Association, Health Information Privacy
Code (1998).
 Corporate bylaws:
 Hospital policies and procedures.
 Municipal Information Acts.
 Best Practices
 COACH Guidelines for the Protection of Health Information.
Sources: OCAP
 Ownership:
 information is owned collectively by the Nation.
 Control:
 the Nation retains control over all aspects of
information management.
 Access:
 the Nation has a right to manage and make decisions
regarding access to their collective information.
 Possession:
 a mechanism to assert ownership.
The inter-provincial view:
Interoperability:
Some Issues:
 Custodians disclosing PHI are generally under a duty to ensure
that the receiving jurisdiction has ‘comparable safeguards’.
 Patients may issue consent directives. Ontario imposes a ‘duty
to notify’ receiving custodians about these.
 Patients should be able to avail themselves of additional
protections in the new jurisdiction.
 Who now has control of the information?
 Consent directives are also sensitive.
More issues:
 Even if we have a way to solve these issues, one of the
major problems is that laws (etc) are dynamic.
Challenge:
 How do we manage policies in a multi-EHR setting?
 Traditional route has been to either purchase COTS
products, or to develop systems for a particular
jurisdiction. (Hard coded business rules).
CHI’s Consent Directives Management System
 Applies constraints prior to providing access or
transmitting PHI.
 Allows consent directives at various levels of
granularity.
 Relies on common privacy vocabulary to apply
consent requirements.
 Can store with EHRi data, or in consolidated form.
Processing Consent Directives in a Jurisdiction
Transfer consent directives from clinical applications
to the EHR.
2. Let either the EHR or (sending clinical application)
process consent directives prior to disclosing a
patient’s PHI.
3. Transfer consent directives from EHR to clinical
applications whenever PHI is disclosed from the
EHR.
Want to avoid having too many consent directives
management systems.
1.
Interjurisdictional Transfer
 Consent directives will be processed whether an access
request is received from a POS system, or clinical
portal, or from an EHR in another jurisdiction.
 Jurisdictions need to agree upon and set policies as to
how consent directives made in one jurisdiction will be
managed following disclosure to another.
 A nationally adopted messaging schema is required for
conveying consent directives between jurisdictions.
Interjurisdictional Transfer (2)
Several goals must be achieved before policy
enforcement can be automated by a policy
management service:
 Jurisdictional policies must be harmonized.
 Rules must be captured and codified.
 Special support for changes to rules.
 Common vocabultary.
Data containing consent directives may flow from one
jurisdiction to another, but policy related data does
not.
Can we do better?
 The inter-jurisdictional data transfer problem is
complex.
 Can we bring some technical tools to bear on the
problem?
 Representing policy rules.
 Operationalizing the representations.
 Storing and securing the representations.
 Managing the representations through their lifecycle.
 Verification and validation.
Current work:
 There has been quite a bit of work on representing policies
and regulations.
 L.Cranor, M. Langehreich, M. Marchiori, J. Reagle, The
Platform for Privacy Preferences (P3P 1.0) Specification.
 R. Agrawal, J. Kiernan, R. Srikant, Y. Xu, An Xpath based
preference language for P3P.
 N. Li, T. Yu, A.I. Anton, A semantics based approach to
privacy languages. (2006)
Current Work
 P. Ashley, S. Hada, G. Karjoth, C. Powers, M. Schunter,
Enterprise Privacy Authorization Language (EPAL 1.1).
 A. Barth, J.C. Mitchell, J. Rosenstein, Conflict and
combination in privacy policy languages (2004).
(DPAL)
 eXtensible Access Control Markup Language.
(XACML)
Current Work
 The above frameworks provide a formalism to specify
data protection policy. They provide methods for
evaluating and enforcing policies.
 Drawback: they are built to manage policies within
single organizations. (Guarda, Zannone, Toward the
Development of Privacy Aware Systems, 2008)
Current Work
 Recent efforts:
 Extend XACML with algorithms addressing issue of policy
similarities and integration across organizations. (Mazzoleni
et al, XACML policy integration algorithms, 2008).
 Distributed temporal logic. (Hilty et al, On obligations, 2005).
 Privacy in Peer to Peer Networks. Automated policy
enforcement. (Weber, Obry).