trafficlight CDA to manage scope and support document evolution

Managing scope is a key part of delivering a successful project - and this is particularly true when using information standards in healthcare projects.  The challenge is that the standards are designed to meet a wide range of requirements, and the project's success will be measured against a specific set of local requirements.

Green CDA was envisaged as a way to take the full CDA specification, and reduce the complexity for implementers by exposing just the information items that matter to the project.  These information items could then either be exposed to the developer as an API, with full CDA documents being generated and shared "over the wire", or could be serialised as an XML document that just contains the Green CDA elements.

One of the problems with Green CDA is that it does not do enough - it tells you the information items that matter to the project - but it does not tell you what to do with the rest of the information items.

TrafficLight CDA is a way to specify more.

• Green for the information items that matter in the project, and that receivers are expected to process or knowingly deal with.  The sender can reasonably expect the receiver to have dealt with this data.
• White for information items that may be ignored or processed by the receiver.  The Sender cannot assume that the receiver will process these, but nor can the sender or receiver expect that the receiving user will see this information.
• Yellow for information that the receiving system must log as out of scope, but which does not need to be brought to the attention of a user before the document is processed or viewed.  The logs must be available for review by the receiving organisation.
• Amber for information that is out of scope, and the inclusion of which should generate a warning which has to be acknowledged by the receiver before the document is processed.  
• Red for information items that must result in the document being rejected, and the sender being informed that the document has not been accepted.

The current CDA specification implies that anything that is not Green or explicitly excluded is White and may be safely ignored.  This proposal builds on existing conformance rules that are described in the HL7 core principles and  in the conformance documentation - an introduces a delightful visual metaphor to help with understanding the different conformance requirements.
Adoption of TrafficLight specifications for how CDA documents are handled will make it easier for suppliers and those testing systems to know what they have to do - and more importantly will ensure that users can clearly understand what will happen when they share a document.

This can be done is a very light-weight way -- saying that anything that is not Green is White (open templates) or saying that anything that is not Green is Red (closed templates).    In practice using the TrafficLights will help projects retain flexibility in an affordable way - enabling explicit distributed evolution of the document set that is supported within a configuration management framework - but that is for another post...

Self monitoring vs telehealth

 Excellent comment my Mary Hawkins on the EHI article.  There is a lack of evidence for most specific implementations of telehealth.  This could be because, as Mary suggests, self-monitoring already happens without devices being directly linked to the EHR or health services - but with the patient or local carer doing the monitoring and calling the health services when needed.

It may be that the technology is moving on faster than the evidence base can keep up with, so there is little incentive to invest in long-term research projects to evaluate technology and methods that will be out of date when the research results are out.

It could also be that we will not call it "telehealth" when it works.  Thousands of people are using online weight monitoring and calorie counting tools - these could be linked up to provide extracts to the NHS.  Gym equipment and pedometers provide electronic logs of the exercise that people are doing in a way that could be linked to EHR data - either for routine care, or for better targeting of care. 

Maybe the way to get "telehealth" working is not to look at buying and installing new kit -- but looking at the way that existing kit and information is being used, and seeing how services can be adapted to make better use of the existing infromation landscape.  Helping patient to become "expert patients" and use the data and data processing tools that they have, and helping healthcare delivery services to be more proactive in finding and reaching out to those who will benefit from their attention, rather than waiting for patients to make appointments.    This will accelerate improvements in health outcomes, and will result in a better definition of what useful telecare services might look like.

Monitoring devices and the "Tele" bit: http://www.ehi.co.uk/news/EHI/8375/pathfinders-losing-their-way

Information Architecture, Archigram, and the Olympics

The Olympics venues in London have been built with reuse in mind, building on the ideas of the Archigram architects, the structures are modular, can be reconfigured, repurposed, or relocated with relative ease (relative being an important word here).  There is a delightful BBC piece describing this [1].
Large Health Information System architectures need a similar flexibility and usability. Olympic buildings take upto seven years to plan and build, and have to find a new purpose after only 4 weeks of use.  For healthcare information systems the planning and building should be shorter than that, but there is not even an initial four weeks that can be relied on where they will be needed and used in the way that was anticipated when planning and building began.
Like the very successful planners for the London Olympics, Healthcare Informaticians could do worse than reread some of the Archigram literature.
[1] http://www.bbc.co.uk/news/entertainment-arts-18811200
[2] http://en.wikipedia.org/wiki/Archigram

Steps to open healthcare data - lets get open specifciations and examples first

Open access to interface specifications and example data for all healthcare IT systems funded by the NHS would enable innovation and quality improvement in the health service.  It is a cheap and effective policy option, aligned with current politics. 

It is understandable that healthcare data such as prescriptions and extracts of medical records are not routinely published for anyone to see.  We all know that this information is private, and belongs to the patient, with the clinicians that recorded it, and the organisations that they work for having some rights and responsibilities to use, take care of, and make available the data  the data when appropriate.

It is understandable that "cleaning" real data so that the patients cannot be identified is hard, and that to do so reliably enough to be comfortable publishing the clean data openly on the internet would be expensive.

What is less understandable is that the specifications for the formats that data may be made available, and invented examples for fictional patients are not available.   The specifications exist but are often only shared once a contract between suppliers and a provider organisation is signed.  They may be based on standards from one of the many SDOs (for which there are rarely examples available), but that is not well published.

Access to specifications and example data would allow new suppliers to show what they could do, and would allow standards organisations (including those setting standards within the NHS) to identify and promote common approaches where there is evident need.  It will allow new entrants to show innovative solutions without requiring "rip-and-replace" for existing systems that deliver value.    
It is reasonable that access to healthcare data should be controlled on behalf of the patient -- but open access to the specifications and example data will help to create  a vibrant marketplace in healthcare IT systems - which is in the interests of patients, clinicians, the NHS and that UK economy.


The Following advice to DHID from an Ehealth Insider  editorial  in May 2012 puts it very well: 

"First, it [DHID] should re-procure the spine as a platform, or series of core re-usable national services, these should be as easy to access and connect to with clear certification criteria. Currently it can take a ludicrous 18 months to go through a Byzantine assurance process. Apple publishes its criteria and manages a rigorous approvals process in less than a month.

The second step would be to insist that all vendors who want to be funded by the NHS meet very basic minimum data standards. Yes, the NHS Commissioning Board is to look at standards, but giving a clear recommendation would avoid re-inventing the wheel. The third step would be to insist that all vendors publish clear, usable Application Programming Interfaces. No API, no funds."

The next step would be to establish open statements saying what has to be done to connect to these interfaces - including the certification criteria and costs.  But getting from where we are to a vibrant market will not be done in one step, so get the APIs published first so that people can create viable and valuable products, then simplify the environment for buying and selling them.

convergence of formal standards and search engine optimization

Formal healthcare information standards organisations are always working to engage effectively with the communities that they serve, and also to discover where their standards are being used, and what issues are being encountered.  Standards that work and meet a need are successful, but it is often hard to discover how to improve specifications that have not got traction. 

Standards Organisations excel at establishing clear definitions, and providing frameworks and specifications against which conformance can be asserted, and which can be included by reference in contracts.  Crucially they also provide a forum where those specifications can be discussed and issues resolved at arms length from the contracts that depend upon them.

Search Engine Optimization (SEO) is the rapidly growing profession of presenting data on the web in a way that results in appropriate placement in search results.  Here there is no problem with engagement - those following "best practice" get to the top for the search terms that they care about.  The process is policed by a combination of published rules and private algorithms, both controlled by commercial search engine organisations, who are in turn motivated (controlled) by commercial drivers to attract and retain users, advertisers, and other revenue streams.

While SEO practitioners do not care about the precise definition of the terms that they are optimising for, they do care about which terms effectively express the relevance of their sites, and which terms the potential users and customers of that site will use.  There is a virtuous circle here of users looking for information that meets their needs, site maintainers making their sites easier to find, policed by the search engine companies and their algorithms who are looking to maximise successful search experiences for their customers.  The search engine companies provide tools to site maintainers to help them improve the relevant ranking of sites - and the site maintainers have business reasons to use these.  There is not the same strong feedback loop in the development of formal standards.

Similar searches are done within organisations using business intelligence engines to extract management and marketing information from the raw data that the organisation holds or has access to.  Here some of the data may have been created for human readers, but much will also have been created conforming to specifications or standards that enable machine processing, albeit not necessarily for the purposes of business intelligence.  This is often transactional data, and changing it requires changes to the workflow processes and the information systems that support them - which is very different to changing the way that websites are structured and populated.  However those of us interested in translational research and innovation in healthcare need to learn from the drivers that are effecting large-scale change in the way that the web looks and functions.

Childish Simplicity

The recent post by Keith Boone suggesting that healthcare information topics should be taught in schools is part of a wider recognition that healthcare IT is far too insular.  We certainly do need to make healthcare information topics available in schools - partly because its useful stuff to understand, and partly because its interesting and challenging.
Another way of doing this that I have been looking at is joining healthcare IT project outreach activities with providing material for existing school courses.  The idea would be to provide "background" material for exercises and coursework that could be used in existing lessons.  What is clear is that the benefits would flow in both directions - more interesting materials for schools, and better public engagement by the healthcare services.  The background material may not be used directly in the classroom, but could be taken home ans shared with the family - this is a well established way to get effective public information out.

Cleaning up the HIT environment: Reduce, Reuse, Recycle

"Reduce, Reuse, Recycle" is a mantra that works for the environment, but how does it play out in the healthcare IT ecosystem? 

Reduce: Stopping the routine collection of needless information has resulted in significant savings in cost, time and frustration.  This requires active review of information flows and uses, but is a very low cost and low risk way to improve things.  It also creates the goodwill and confidence that will sustain other change initiatives.   Focusing on specific Collaborations where information needs to be shared for a particular purpose is also a way to avoid the collection of precise data when something simpler is good enough.
Reuse: Finding ways to reuse existing information systems and infrastructure rather than assuming that it must be replaced makes sense.  This applies to software and networks, but it also applies to paper forms and processes.  Using architectural frameworks and standards-based interfaces to support incremental change, and a collaboration-based strategy that focuses on the human and systems issues that support specific collaborations allows for strategic improvements to be introduced with minimal change.  
Change management is hard - leveraging the reuse of infrastructure, processes, skills and ideas is a good way to increase the benefit:change ratio - and so the return on expensive and risky change management effort.
Recycle:  Healthcare data is full of "toxic" confidential information that prevents it being made available for purposes other that those for which it was collected.  However we get value out of scrap metal and glass by dramatically denaturing it and throwing away much of the structure that was painstakingly designed in for the initial use.  We should be getting value out of de-identified healthcare data.   Some recycling is better than none - so even if much of the semantic value has to be lost to avoid perceived or actual risks to confidentiality - in the interests of health we should be recycling more.  The next step will be to make our healthcare data more recyclable in the first place, just as has been done with physical packaging.  How to promote recyclable healthcare information is a topic for another day.

There is nothing new or radical in this post - other than reusing an idea from one area where widescale and dramatic change is needed (the environment), and applying it to another (health).