In this post, I muse on Denmark’s implementation of a country-wide system for secure, up-to-date sharing of EMRs and patient identity federation. But I primarily want to share a links for those interested in what they are doing:
A Cute Introduction
Last week Barack Obama visited Copenhagen to support his home city’s bid to have the 2016 Olympics hosted in Chicago. Later this year the U.S. President will meet with international leaders in Copenhagen for a UN summit, negotiating the successor to the Kyoto protocol.
In U.S. political news, the international happenings in Denmark have offered a nice break from the ongoing, rancorous national debate over reforming the U.S. health care system. Political events have stirred a broader conversation about the overall state of American health care, such as the cost and effectiveness of the current system. In a moment of free association, the events in Denmark reminded me of some interesting things about that nation’s health care system: the Danes are rather progressive—no, not because they’ve socialized, I’ll entirely leave this matter aside—in regards to they’re health care IT infrastructure.
What is Denmark Doing?
Denmark’s system is interesting so I’ll share what I’ve learned of the nation’s overall approach to health care IT and, in greater detail, discuss their implementation of PKI.
There are many Danish organizations involved with the reform of health care IT. Foremost are MedCom, the Danish Centre for Health Telematics, who is the coordinating organization for health care in Denmark and manager of the Danish Health Data Network; the National Board of Health for Denmark, who developed the data model and terminology server for the system, and leads the country’s overall health IT stragegy; and the Ministry of Science, Technology and Innovation (MTVU) in Denmark, who develops most of Denmark’s technical standards and recommended a standard for Service-Oriented Architecture (SOA) identity federation to be used in various Danish systems.
The National Board of Health’s stated goal for the reformed system was “to provide a connected health care sector in which health professionals have access to all relevant EHR data regardless of where citizens seek treatment and no matter where or when this information was registered.” Lofty, indeed 1. Unlike most countries, though, Denmark has robust broadband access in most of the country. And most general practices and hospitals already use electronic medical records (EMRs). The National Board of Health knew it would need to implement a nationwide SOA for the secure web sharing of data.
Implementation of PKI
Denmark built it’s PKI on top of it’s existing virtual private network (VPN) architecture, which is made available to all health care providers in the country, and it was already in use by many for remote collaboration. At the behest of MVTU, SAML was selected as the framework for identity federation and the exchange of authentication assertions. Health care professionals are issued DanID, a X.509 certificate from the Danish OECS CA. The following step explain how authentication is performed between Danish health systems 2:
- User authenticates as part of login to local EHR system and a digitally signed, SAML assertion is created.
– this is a SAML security token, referred to as a virtual health professional identity card.
- A direct request is made to a central security token service (STS), which checks the validity of the local system’s digital signature, the user’s signature, certificate validity and revocation status, and core certificate attributes3.
- STS signs the SAML token and sends a response to the local system.
- The SAML security token can be used until it expires (after 24 hours).
I’m not sure what plans Denmark has for the authentication of everyday citizens to health care services and portals4. The foundations are certainly in place. The infrastructure for the clinical exchange of medical records, which utilizes the Danish Central Person Registry (number), provides a unique identifier for all national patients. Sundhed.dk is a public portal for Danish citizens where patients can access (some) of their health information, receive online consultation, schedule health services, and renew prescriptions/treatments. While Denmark does not issue electronic ID cards, each citizen is given a digital certificate, which is automatically derived from that citizen’s CPR number. With a combination of these parts, each Danish citizen could use their digital certificate for authentication to sundhed.dk and for the signing of health documents.
Lesson from Denmark’s System?
What can be learned from Denmark? Well, one could try to point out the things Denmark has done right, as Gartner did in their study, which will be either unmissable or made up: Denmark used a “[g]radual approach with realistic time frames”; they gave “Incentives to vendors”; they used a “project-based approach”; they “[kept] an appropriate balance between central coordination and local leadership.”; the country has a “culture of consensus”.
As all observers have pointed out, its too early to tell what improvements the reformed IT changes have made. What Denmark seems to have done right is to start with a basic, but sound architecture that makes use of existing infrastructure and technologies. They have similarly, worked to make the systems simple, affordable, and feasible for all of the country’s health providers, using open standards and technologies.
Beyond the broader success of the program, I was interested to understand how adoption and use of the PKI has been. But, it seem too early to ascertain the problems with the reformed system or understand the parts of the systems that will need to be improved. From TrustBearer’s perspective, we are interested in problems experienced while deploying and using PKI, issues such as interoperability between relying systems, certificate policies, certificate validation, and renewal, distinguishing between levels of identity assurance, and usability for end-users. I could not find much information in regard to these issues in the Danish system, so this will be a topic left for future blog posts. One thing of note was that developers involved in the Danish project found some things lacking in the the SAML/XML schema, because its was not possible to express certain types of requirements and policies as part of an authentication/authorization assertion5. (This is related, rather loosely, to a problem TrustBearer was trying to solve in another context, signifying the strength of an authentication method in the OpenID Provider Authentication Policy Extension.)
1. A Federation of Web Services for Danish Health Care
2. As outlined in A Federation of Web Services for Danish Health Care.
3. Exchange of tokens over SOAP. http://docs.oasis-open.org/ws-sx/ws-trust/v1.3/ws-trust.html
4. There is a least one pilot of software-certificate-based PKI access for out patients. http://nortelemed.custompublish.com/healthnets-and-new-services-electronic-patient-records.47966-5213.html?id=48665&cat=5385