A lot of activity and progress is underway around the world right now, and has been for some time, regarding integrating and sharing health data for healthcare management and delivery purposes. Many standards, reference models and authorities have arisen to guide implementation and use of IT for these purposes, for example health information exchange standards driven by the Office of the National Coordinator for Health Information Technology (ONC – http://www.healthit.gov/). Many very new and modern health IT capabilities and products are available now, alongside systems and data that may have been first created over 30 years ago (particularly in the Federal Government).
In the media and within procurement activity, the swirl of misused phrases and definitions isn’t clarifying many approaches. Records vs. Data vs. Information. Interoperability vs. Integration. Standards vs. Policies. Systems vs. Software vs. Products or Solutions. COTS vs. Services vs. Modules vs. Applications. Open Source vs. Open Standards. Modern vs. Legacy vs. Current.
In Enterprise Architecture (EA) terms, the messages regarding Integrated Healthcare IT requirements aren’t commonly being presented at a consistent level of abstraction, according to a consistent architecture model and vocabulary. As well, the audience or consumers of this information aren’t being addressed in ways most impactful to their specific needs and concerns.
What are the audience concerns? IT system owners need to maintain data security and system performance, within technology and investment constraints. Doctors need consistent, instant, reliable and comprehensive visualization of data and the point of care. Government oversight bodies need recurring validation that money is spent wisely and results meet both mission and legislative requirements. Veterans, soldiers and their families need absolutely private, accurate, real-time information about their healthcare status – wherever they are. The pharmaceutical and medical device industries need timely, useful data regarding outcomes and utilization – to drive product improvement and cost-effectiveness. Hospitals, clinics and transport services need utilization and clinical workflow measurements to manage personnel and equipment resources.
The highest separation of concerns can be segmented by standard Enterprise Architecture domains or “views”. A very generic, traditional model is the “BAIT” model – i.e. Business, Application, Information and Technology. Note that this is very similar to the widely-known “ISO Reference Model for Open Distributed Processing” (RM-ODP) Viewpoints – which underpin evolving healthcare standards including the “HL7 Services Aware Interoperability Framework” (SAIF).
The “Business Domain” encompasses the discussion about business processes, financials, resources and logistics, organization and roles. Who does what, under what circumstances or authority, and how outcomes are evaluated and purchased. The business drivers and enablers of successful healthcare delivery, one might say.
The “Application Domain” concerns automating the “practice of healthcare”. Automated systems (and their user interfaces) are very helpful in planning, monitoring and managing the workflow, resources and facility environments, and of course processing data for clinical care, surveillance and health data management and reporting purposes. This is where healthcare expertise is codified in software and device configurations, where medical intelligence and knowledge meets computer-enabled automation. This domain is the chief concern of clinical practitioners and patients – where they can most helpfully provide requirements and evaluate results. Software that’s built to process healthcare data comes in many shapes and sizes, can be owned or rented, are proprietary or completely transparent.
The “Information Domain” is in essence the “fuel” for the Application Domain. Healthcare practitioners and patients care that this fuel is reliable, protected and of the highest quality – but aren’t too invested in how this is achieved, beyond required or trained procedures. It’s like filling the car with gas – there’s some choice and control, but fundamentally a lot of trust that the gas will do the job. For those whose concern is actually delivering gas – from undersea oil deposits all the way to the pump – this domain is an industry unto itself. Likewise, collecting, repurposing, sharing, analyzing information about patient and provider healthcare status is a required platform on which successful healthcare user applications and interfaces are built. This is what “Chief Medical Information Officers” are concerned with, as are “Medical Informatics Professionals”. They are also concerned with the difference between healthcare “records”, “archives” and “information” – but that’s a discussion for another day.
It is critical to note that “Information” is composed of data; core or “raw” data is packaged, assembled, standardized, illustrated, modeled and summarized as information more easily consumed and understood by users. Pictures, sound bites and brief notes taken by an officer at an accident scene are data (as are “Big Data” signals from public social media and traffic sensors); the information packages include the accident report, the newspaper article, the insurance claim and the emergency room evaluation. These days, with the proliferation of data-generating devices and sensors, along with the rapid data replication and distribution channels available over the Internet, the “Data Domain” itself can be a nearly independent concern of some – providing the raw fuel to the information fire, oil for refined gas.
The “Technology Domain” is essentially all of the electronic computing and data storage elements needed to manage data and resulting information, operate software and deliver the software results to user interfaces (like browsers, video screens, medical devices). Things like servers, mobile phones, physical sensors, telecommunications networks, storage repositories – this includes the machine-specific software embedded into medical equipment.
Sidebar: Data Domain Standards
Quite a bit of work and investment is required to collect, filter, store, protect and make available raw data across the clinical care lifecycle, in order that the right kind of information is then available to be utilized by users or software. Most importantly, reusable, open standards and Reference Implementation Models (RIMs) concerned with the Data Management domain are foundation requirements for any effective healthcare information system that participates in the global healthcare ecosystem.
A RIM is basically working software or implementation patterns for testing and confirming compliance with standards, thereby promoting creation of software products that incorporate and maintain the standards. It’s a reusable, implementable, working set of code with documentation – focused on a common concern, decoupled from implementation policies or constraints. RIMs are useful for facilitating standards adoption across collaborative software development communities at every layer of the Enterprise Architecture.
For example, a data-domain RIM developed several years ago by Oracle Health Sciences (in a clinical research setting) focused on maintaining role-based access security requirements when two existing sets of research and patient care data were merged for querying. The design of the single RIM merged the HL7 Clinical Research Model (BRIDG) with an HL7 EHR Model (Care Record) to support a working proof-of-concept – that others could adopt as relevant. The “concern” here was data security – separate from the information and application-level concerns of enabling multi-repository information visualization methods for researchers.
The point of this discussion on EA-driven separation of concerns is illustrated as follows. When a spokesman (or RFP author) says “the system will be interoperable” – it’s likely that by “system” the meaning is some segment of the “Application Domain” being able to exchange objects from the “Information Domain”. Instead, a better phrase might be “the software application will be able to share standardized healthcare information with other applications”. This keeps the principle discussion at the application and information-sharing software level, and doesn’t make detailed assumptions or predictions regarding concerns in the Business, Data or Technology Domains. Those are different, but related discussions, and may already be addressed by reusable, standard offerings, RIMs or acquisition strategies.
Taking this approach to broadly interpret the recent announcement that the DoD will seek a competitive procurement for “Healthcare Management Software Modernization” – it appears the focus of this need is the Application Domain – i.e. software packages and/or services that generate and use healthcare information while managing healthcare processes and interactions.
To support these new software application features, separate but related activity is required to address “modernization” concerns among the other EA domains – concerns relating to datacenter infrastructure, data management and security services, end-user devices and interfaces, etc. Some of this activity may not be dedicated to healthcare management, but be shared and supported for enterprise use, for other missions. That’s why use of a current, relevant EA frameworks (such as DODAF v2.02 and the OMB “Common Approach” ) is so important, managing shared capabilities and investments.
Using standard EA viewpoints to separate concerns will also expose reuse opportunities (and possibly consolidate or reduce acquisition needs), i.e. leveraging existing investments that are practical enablers. Some examples might include the developing iEHR health record structured message translation and sharing services, plus HHS/ONC initiatives including Health Information Exchange Networks and the “VA Blue Button” personal health record service.
Back in 2008, I wrote a paper on Business Analysis. Recently, I’ve been revisiting this subject in my day-job and this made me realise how little things had changed and inspired me to write this post, which is, basically, the original article re-written, with additional thoughts (in italics).
