Modernizing VistA: Enabling scalable interoperability in a legacy system
- By Mark Byers
- Sep 11, 2014
The Department of Veterans Affairs is in the midst of an evolution as it continues efforts to modernize and ensure the long-term viability of VistA to meet the healthcare needs of the nation’s military veterans.
The Veterans Health Information Systems and Technology Architecture is the VA’s enterprise electronic health record system that tracks clinical treatment and care information.
Upgrading it is a major undertaking with a significant impact on healthcare facilities across the VA, from both clinical and technology standpoints.
While the technical challenges are formidable, leveraging cutting-edge tools and platforms can be part of the solution for achieving fast, secure and scalable interoperability within the mission-critical legacy system. However, the needs of clinical staff must remain paramount in designing solutions. A seemingly elegant systems architecture, for instance, is not viable unless it enables maximum efficiency for the providers who use it to treat the nation’s veterans. Thus, two technology approaches could have substantial effects on performance.
Approach 1: Tackling the abstraction layer dilemma
The overall modernization strategy to date has focused on placing abstraction layers between VistA and newer software components. In theory, this is a sound approach when working with a legacy system. In practice, however, this architectural layering often comes at the expense of speed and efficiency for solutions and ultimately end users.
Providers care about the number of mouse clicks required, the total seconds spent waiting and the number of applications they must touch to access patient information related to electronic health records. While a few additional clicks here and there or a few seconds to refresh a screen may not seem like much, with thousands of providers accessing VistA every hour, the end result is really an unacceptable time lag and contributes to a systemic efficiency disaster.
From an enterprise integration perspective, multiple layers of service oriented (SOA) middleware and service busses add both latency and bloat to a legacy integration solution. The latency itself should be enough to render such a system ‘unusable’ from the real-world perspective of, say, web users waiting on a legacy-integrated web app to perform meaningful work. However, such a solution also multiplies the number of legacy integration points, adding multiple points of failure and unnecessary complexity.
Previous attempts to place a Simple Object Access Protocol (SOAP) layer in front of VistA has only provided developers with ‘a different way to access the same data,’ which has the effect of making the overall system slower and less responsive.
In addition, layering a transport solution (SOAP) on top of a business logic problem (ex: collating lab results) contributes nothing toward the much needed centralization of business logic in the first place. This means that, for example, developers requesting the same legacy data, regardless of how these data are actually published and retrieved (SOA, legacy remote procedure calls (RPC) or otherwise), could end up solving the same business logic problems myriad times.
One solution is to move all business logic to the middle tier using a platform-based approach. In a sense, such a solution would “right size” the problem by moving both the service layer up and the application business logic down to the same middle-tier location.
Deploying the business logic in a central location also makes it enterprise-accessible from any device or location. This solution extends and exposes all of the VistA business logic RPCs into the middle tier as SOAP/JSON-accessible operations across multiple VistA instances. This provides developers with a truly read-write service interface layer, allowing them to integrate VistA data into all modern web-, mobile- and cloud-based solutions.
A VistA-integrated deployment platform and engine for business logic with the ability to establish and query VistA across an authenticated pipe represents a more efficient approach for making authenticated, user-specific, user-identified calls to remote systems.
Further, by using authenticated connection pooling, it provides a fast and highly secure solution. From a clinical perspective, a multisite, authenticated, context-sensitive user access solution also allows VA staff to aggregate regional or national data into much needed patient-centric longitudinal records.
Approach 2: Standardizing HL7 using an integration framework
From order entry to labs to radiology and more, some messages generated within VistA are not true Health Level-7 (HL7) standard messages, but rather message stubs that are simply passed between applications. HL7 messages need to be standardized across VistA packages for usability by commercial off-the-shelf vendors and machine interfaces. This includes using common delimiters and standardizing the use and placement of patient identifiers and demographic fields as well as following HL7 conformance rules for message segments and fields.
This challenge could be solved with a commercial interface engine or customized parsers. Although one major drawback with this approach across a large, distributed, high-volume enterprise is that it might require a massive amount of coordinated script writing, some solutions such as a HL7 integration framework eliminate the need for scripting.
Sitting on top of the VistA HL7 package, such a framework provides the capability for standardization and screening of messages to customize output and reduce network traffic while supporting VistA HL7 dynamic addressing – where the endpoint is not selected until the message is generated. This eliminates the need to define thousands of protocols within VistA when only a single, standardized protocol is required.
In tandem, an enterprise integration service and a HL7 integration framework address several of the most significant technical challenges faced by the VA in modernizing VistA. By reducing the amount of time spent collecting and correlating patient data, solutions such as these can create faster and more efficient workflows and ultimately better care for the nation’s veterans.
Mark Byers is president and CEO of DSS. His experience spans more than 33 years in the data processing industry, and he has been involved in the development of electronic health record software, specifically the VistA-based platform, for more than 16 years. Follow him on Twitter: @DSSHealthIT.