My output fulfils what I was looking to design as set out in my brief. However, it is not without downsides.
As discussed in distribution, Health Institutions have to weight the cost + risk of upgrading their software against the potential value provided by the new system - there is a socio-technical impact. Barnett's approach to this is to reduce the risk, and cost of upgrading software, with it's extensible, iterative design. Changes can be implemented with minimal disruption or risk. However, what it doesn't solve is how to overcome this problem for the first time implementation.
As discussed in "Electronic health records: research into design and implementation", implementing an EHR system is a large sociotechnical change (Beasley, Holden, & Sullivan, 2011). EHR systems are not simply cleanly added onto existing systems, they require a full transformation of not only the software, by workflows and processes used in healthcare institutions. The impact of implementing such system is not only technical in nature, but societal.
While Barnett provides a clean way to iterate upon itself with minimal impact, its does not address how it could be cleanly built upon, or perfectly replace, existing systems. The initial implementation would still be a huge socio-technical change for an institution. Researching how an institution could make this transition easier was simply out of scope for this thesis. It would require a close, long-term working relationship with a large range of institutions.
However, this is an area for further research, for both Barnett and EHR systems in general - "How can institutions make smoother migrations to EHR systems?".
Related is the second issue - the cost. Developing a system like Barnett would take a lot of development time and testing. In order for a primary care institution to switch to Barnett, they also require additional functionality, such as appointment management. Barnett was designed with the intention that separate software and services would provide that functionality, due to the advantages of compartmentalising complexity. However, for an institution to use this system, that functionality must exist, so these services must also be created.
This of course, increases the cost by a lot. The question then is, is it worth it? In distribution, I question the value provided by better software. While there are undoubtedly benefits to EHR systems, when the cost can easily hit several billion dollars, is it worth it? Will it really have the transformative impact on healthcare that you'd expect from a billion dollar budget?
I think long term a system like Barnett would reduce the cost. However, the upfront spending is high, and it's unclear what party should bear the cost - institutions, patients, or the government?
There is also a potential for a critique of methods. Finding medical professionals willing to spend their precious time to interview with me, was a challenge. In the end, I found four out of six of my candidates from a GP IT research group. This research group was a group of GPs who had self-selected an interest in Health IT.
This could potentially create bias - the individuals I talked were clearly passionate about Health IT, and because of that their opinions cannot be seen as a fully truthful representation of GPs throughout New Zealand. They might have been much more dissatisfied with their tools than other GPs, and had much grander expectations of what their software should do.
After completing this thesis, I believe there are three potential futures for Health IT.
1 - Nothing changes. Software is continued to be made for healthcare in exactly the same fashion - with disappointing outcomes, slow production speed, and high cost.
2 - Healthcare software is made with better processes. The industry as a whole starts improving the methodologies around building healthcare software, and thinks long term. For example, complex systems are broken into smaller systems, vendor lock-in is reduced, and faster iteration loops are set up.
3 - A leap frog technology. A technology comes along which provides either incredible value, or can be implemented with little cost or risk. Modern Machine Learning and Artificial Intelligence have the potential to be this technology, but there are huge challenges to this becoming a reality.
Barnett is firmly centred in the second future - a future where we're not leveraging huge budgets, or futuristic technology, but collectively building with a common goal and methodology of iterative improvement. What future will succeed? Only time can tell.
By Eliot Slevin
eliot.slevin@gmail.com