To get national healthcare right requires Adaptive Intelligence
With the employer mandate delays being the latest setback to U.S. president Obama's push for national healthcare, it's worth looking at how other countries are successfully tackling the same problem. The United Kingdom has had nationalized healthcare for years, and one of the things that makes this effort so successful is its approach to data collaboration — something Forrester calls Adaptive Intelligence.
While the UK hasn't successfully moved into fully electronic health records, it has in place today a health records sharing system that lets its over 27,000 member organizations string together patient care information across providers, hospitals, and ministries, creating a more full and accurate picture of each patient, which results in better care. At the heart of this exchange is a central data sharing system called Spine. It's through Spine that all the National Health Service (NHS) member organizations connect their data sets for integration and analysis. The data-sharing model Spine creates has been integral in the creation of summary care records across providers, an electronic prescription service, and highly detailed patient care quality analysis. As we discussed in the Forrester report "Introducing Adaptive Intelligence," no one company can alone create an accurate picture of its customers or its business without collaborating on the data and analysis with other organizations who have complementary views that flesh out the picture.
Forrester defines Adaptive Intelligence as:
Real-time, multi-directional sharing of data to derive contextually appropriate, authoritative knowledge that helps maximize business value.
And this value is exactly what Spine provides for the NHS.
Spine, however, is coming to the end of its life. It's based on a traditional scale up, three-tiered architecture with an Oracle database at the backend. Being accessed by so many organizations — and being part of so many technology-supported business processes — availability, change costs, and response times associated with Spine are critical to the NHS. As such, the NHS is deep into the development and testing of Spine 2, which moves the core data sharing system to a modern application architecture that is much more flexible and agile. The front and middle tiers are built on mostly open source technologies including RabbitMQ, Redis, Tornado, Splunk, and others. The backend relational database is being replaced by a NoSQL key value store using Riak from Basho Technologies, while the front tiers will maintain existing XML-based service connections and integrations. The core reason the NHS chose this new architecture was cost and performance requirements.
“We’re looking for a 90 percent reduction in response times for key messages and a radical improvement in the cost of change,” said Martin Sumner, the technical lead on the Spine2 project.
Sumner added that the architectural shift and the broader change in approach to using in-house expertise, supported by subject matter experts (SMEs) with Agile techniques, had reduced the core technical team to less than 30 people. This is a tremendous improvement over Spine 1, which the incumbent supplier considered to be one of the largest IT programs in the world, "consuming over 15,000 man-years of effort to date."
Core to any effective Adaptive Intelligence sharing project is good data quality management, which requires, in a healthcare context, consistent patient identifiers across the system. In this case, the NHS, as the data authority, maintains a single NHS number for each UK citizen, which is the personal identifier within the system. One of the keys to performance improvement in Spine 2 is the separation of the system into three services: core real-time data integration services, identity management, and secondary uses. This will let the NHS scale and manage the availability of each service separately — another key best practice in modern application design.
Unlike the current Spine, the next version will give the NHS performance efficiency through horizontal scaling, a more synchronous messaging interface, innovative records management, and a modern, efficient storage system. The data set isn’t huge — Sumner said Spine is in the low TBs in size today — but is widely used and at the heart of an integration culture that has grown around the model of Adaptive Intelligence.
So while the U.S. Congress is kept busy debating the timing for its move to national healthcare, the IT side of the government might want to take a look at the data sharing model it will need to put in place once a date is finally agreed upon.